diff --git a/Predict-Lung-Disease-master/000_preprocess.py b/Predict-Lung-Disease-master/000_preprocess.py
new file mode 100644
index 0000000..78e182f
--- /dev/null
+++ b/Predict-Lung-Disease-master/000_preprocess.py
@@ -0,0 +1,164 @@
+import sys, os
+import azure_chestxray_utils
+import pickle
+import random
+import re
+import tqdm
+import cv2
+import numpy as np
+import pandas as pd
+import sklearn.model_selection
+from collections import Counter
+
+paths_to_append = [os.path.join(os.getcwd(), os.path.join(*(['Code',  'src'])))]
+def add_path_to_sys_path(path_to_append):
+    if not (any(path_to_append in paths for paths in sys.path)):
+        sys.path.append(path_to_append)
+
+[add_path_to_sys_path(crt_path) for crt_path in paths_to_append]
+
+path= os.getcwd()+r'\azure-share'
+isExists=os.path.exists(path)
+if not isExists:
+    amlWBSharedDir = os.mkdir(path)
+else:
+    amlWBSharedDir = path
+
+
+
+
+prj_consts = azure_chestxray_utils.chestxray_consts()
+print(prj_consts)
+
+data_base_input_dir=os.path.join(amlWBSharedDir, os.path.join(*(prj_consts.BASE_INPUT_DIR_list)))
+data_base_output_dir=os.path.join(amlWBSharedDir, os.path.join(*(prj_consts.BASE_OUTPUT_DIR_list)))
+
+isExists1 = os.path.exists(data_base_input_dir)
+isExists2 = os.path.exists(data_base_output_dir)
+
+if not isExists1:
+    data_base_input_dir = os.mkdir(data_base_input_dir)
+print(data_base_input_dir)
+
+if not isExists2:
+    data_base_output_dir = os.mkdir(data_base_output_dir)
+print(data_base_output_dir)
+
+nih_chest_xray_data_dir=os.path.join(data_base_input_dir,
+                                     os.path.join(*(prj_consts.ChestXray_IMAGES_DIR_list)))
+isExists3 = os.path.exists(nih_chest_xray_data_dir)
+if not isExists3:
+    nih_chest_xray_data_dir = os.mkdir(nih_chest_xray_data_dir)
+
+print(nih_chest_xray_data_dir)
+
+other_data_dir=os.path.join(data_base_input_dir, os.path.join(*(prj_consts.ChestXray_OTHER_DATA_DIR_list)))
+data_partitions_dir=os.path.join(data_base_output_dir, os.path.join(*(prj_consts.DATA_PARTITIONS_DIR_list)))
+
+ignored_images_set = set()
+
+total_patient_number = 30805
+NIH_annotated_file = 'BBox_List_2017.csv' # exclude from train pathology annotated by radiologists
+manually_selected_bad_images_file = 'blacklist.csv'# exclude what viusally looks like bad images
+
+patient_id_original = [i for i in range(1,total_patient_number + 1)]
+
+bbox_df = pd.read_csv(os.path.join(other_data_dir, NIH_annotated_file))
+bbox_patient_index_df = bbox_df['Image Index'].str.slice(3, 8)
+
+bbox_patient_index_list = []
+for index, item in bbox_patient_index_df.iteritems():
+    bbox_patient_index_list.append(int(item))
+
+patient_id = list(set(patient_id_original) - set(bbox_patient_index_list))
+print("len of original patient id is", len(patient_id_original))
+print("len of cleaned patient id is", len(patient_id))
+print("len of unique patient id with annotated data",
+      len(list(set(bbox_patient_index_list))))
+print("len of patient id with annotated data",bbox_df.shape[0])
+
+random.seed(0)
+random.shuffle(patient_id)
+
+print("first ten patient ids are", patient_id[:10])
+
+# training:valid:test=7:1:2
+patient_id_train = patient_id[:int(total_patient_number * 0.7)]
+patient_id_valid = patient_id[int(total_patient_number * 0.7):int(total_patient_number * 0.8)]
+# get the rest of the patient_id as the test set
+patient_id_test = patient_id[int(total_patient_number * 0.8):]
+patient_id_test.extend(bbox_patient_index_list)
+patient_id_test = list(set(patient_id_test))
+
+print("train:{} valid:{} test:{}".format(len(patient_id_train), len(patient_id_valid), len(patient_id_test)))
+
+pathologies_name_list = prj_consts.DISEASE_list
+NIH_patients_and_labels_file = 'Data_Entry_2017.csv'
+
+labels_df = pd.read_csv(os.path.join(other_data_dir, NIH_patients_and_labels_file))
+
+
+#show the label distribution
+
+# Unique IDs frequencies can be computed using list comprehension or collections lib
+# [[x,(list(crtData['fullID2'])).count(x)] for x in set(crtData['fullID2'])]
+# for tallying, collections lib is faster than list comprehension
+pathology_distribution = Counter(list(labels_df['Finding Labels']))
+
+# Sort it by ID frequency (dict value)
+sorted_by_freq = sorted(pathology_distribution.items(), key=lambda x: x[1], reverse=True)
+print(len(sorted_by_freq))
+print(sorted_by_freq[:20])
+print(sorted_by_freq[-10:])
+
+print(labels_df['Finding Labels'].str.split( '|', expand=False).str.join(sep='*').str.get_dummies(sep='*').sum())
+
+def process_data(current_df, patient_ids):
+    image_name_index = []
+    image_labels = {}
+    for individual_patient in tqdm.tqdm(patient_ids):
+        for _, row in current_df[current_df['Patient ID'] == individual_patient].iterrows():
+            processed_image_name = row['Image Index']
+            if processed_image_name in ignored_images_set:
+                pass
+            else:
+                image_name_index.append(processed_image_name)
+                image_labels[processed_image_name] = np.zeros(14, dtype=np.uint8)
+                for disease_index, ele in enumerate(pathologies_name_list):
+                    if re.search(ele, row['Finding Labels'], re.IGNORECASE):
+                        image_labels[processed_image_name][disease_index] = 1
+                    else:
+                        # redundant code but just to make it more readable
+                        image_labels[processed_image_name][disease_index] = 0
+                # print("processed", row['Image Index'])
+    return image_name_index, image_labels
+
+
+train_data_index, train_labels = process_data(labels_df, patient_id_train)
+valid_data_index, valid_labels = process_data(labels_df, patient_id_valid)
+test_data_index, test_labels = process_data(labels_df, patient_id_test)
+
+print("train, valid, test image number is:", len(train_data_index), len(valid_data_index), len(test_data_index))
+
+# save the data
+labels_all = {}
+labels_all.update(train_labels)
+labels_all.update(valid_labels)
+labels_all.update(test_labels)
+
+partition_dict = {'train': train_data_index, 'test': test_data_index, 'valid': valid_data_index}
+
+with open(os.path.join(data_partitions_dir, 'labels14_unormalized_cleaned.pickle'), 'wb') as f:
+    pickle.dump(labels_all, f)
+
+with open(os.path.join(data_partitions_dir, 'partition14_unormalized_cleaned.pickle'), 'wb') as f:
+    pickle.dump(partition_dict, f)
+
+# also save the patient id partitions for pytorch training
+with open(os.path.join(data_partitions_dir, 'train_test_valid_data_partitions.pickle'), 'wb') as f:
+    pickle.dump([patient_id_train, patient_id_valid,
+                 patient_id_test,
+                 list(set(bbox_patient_index_list))], f)
+
+print(type(train_labels))
+print({k: train_labels[k] for k in list(train_labels)[:5]})
\ No newline at end of file
diff --git a/Predict-Lung-Disease-master/020_evaluate.py b/Predict-Lung-Disease-master/020_evaluate.py
new file mode 100644
index 0000000..0064c2c
--- /dev/null
+++ b/Predict-Lung-Disease-master/020_evaluate.py
@@ -0,0 +1,157 @@
+import sys, os
+import azure_chestxray_utils
+import azure_chestxray_keras_utils
+from keras.models import load_model
+import os
+import pickle
+import cv2
+import numpy as np
+import pandas as pd
+from keras.models import load_model
+from keras.utils import Sequence
+from sklearn import metrics
+from tensorflow.python.client import device_lib
+import keras_contrib
+
+path = os.getcwd()+r'\azure-share'
+amlWBSharedDir = path
+
+prj_consts = azure_chestxray_utils.chestxray_consts()
+data_base_input_dir=os.path.join(amlWBSharedDir,
+                                 os.path.join(*(prj_consts.BASE_INPUT_DIR_list)))
+data_base_output_dir=os.path.join(amlWBSharedDir,
+                                  os.path.join(*(prj_consts.BASE_OUTPUT_DIR_list)))
+weights_dir = os.path.join(data_base_output_dir, os.path.join(*(prj_consts.MODEL_WEIGHTS_DIR_list)))
+fully_trained_weights_dir = os.path.join(data_base_output_dir, os.path.join(*(prj_consts.FULLY_PRETRAINED_MODEL_DIR_list)))
+
+nih_chest_xray_data_dir = os.path.join(data_base_input_dir,
+                                     os.path.join(*(prj_consts.ChestXray_IMAGES_DIR_list)))
+
+data_partitions_dir = os.path.join(data_base_output_dir,
+                                os.path.join(*(prj_consts.DATA_PARTITIONS_DIR_list)))
+
+label_path = os.path.join(data_partitions_dir,'labels14_unormalized_cleaned.pickle')
+
+partition_path = os.path.join(data_partitions_dir, 'partition14_unormalized_cleaned.pickle')
+
+model_file_name = 'azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5'
+model = load_model(os.path.join(fully_trained_weights_dir, model_file_name))
+model.save_weights(os.path.join(fully_trained_weights_dir, 'weights_only_'+model_file_name))
+models_file_name= [os.path.join(fully_trained_weights_dir,
+                               'weights_only_azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5')]
+os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"  # see issue #152
+os.environ["CUDA_VISIBLE_DEVICES"] = "0,1"
+
+
+
+resized_height = 224
+resized_width = 224
+num_channel = 3
+num_classes = 14
+batch_size = 100 #512
+
+def get_available_gpus():
+    """
+    Returns: number of GPUs available in the system
+    """
+    local_device_protos = device_lib.list_local_devices()
+    return [x.name for x in local_device_protos if x.device_type == 'GPU']
+
+get_available_gpus()
+# get number of available GPUs
+print("num of GPUs:", len(get_available_gpus()))
+
+num_gpu = get_available_gpus()
+# get number of available GPUs
+print("num of GPUs:", len(get_available_gpus()))
+
+pathologies_name_list = prj_consts.DISEASE_list
+pathologies_name_list
+
+stanford_result = [0.8094, 0.9248, 0.8638, 0.7345, 0.8676, 0.7802, 0.7680, 0.8887, 0.7901, 0.8878, 0.9371, 0.8047,
+                   0.8062, 0.9164]
+
+
+with open(label_path, 'rb') as f:
+    labels = pickle.load(f)
+
+with open(partition_path, 'rb') as f:
+    partition = pickle.load(f)
+
+class DataGenSequence(Sequence):
+    def __init__(self, labels, image_file_index, current_state):
+        self.batch_size = batch_size
+        self.labels = labels
+        self.img_file_index = image_file_index
+        self.current_state = current_state
+        self.len = len(self.img_file_index) // self.batch_size
+        print("for DataGenSequence", current_state, "total rows are:", len(self.img_file_index), ", len is", self.len)
+
+    def __len__(self):
+        return self.len
+
+    def __getitem__(self, idx):
+        # print("loading data segmentation", idx)
+        # make sure each batch size has the same amount of data
+        current_batch = self.img_file_index[idx * self.batch_size: (idx + 1) * self.batch_size]
+        X = np.empty((self.batch_size, resized_height, resized_width, num_channel))
+        y = np.empty((self.batch_size, num_classes))
+
+        for i, image_name in enumerate(current_batch):
+            path = os.path.join(nih_chest_xray_data_dir, image_name)
+
+            # loading data
+
+            img = cv2.resize(cv2.imread(path), (resized_height, resized_width)).astype(np.float16)
+            X[i, :, :, :] = img
+            y[i, :] = labels[image_name]
+
+            # only do random flipping in training status
+        if self.current_state == 'train':
+            # this is different from the training code
+            x_augmented = X
+        else:
+            x_augmented = X
+
+        return x_augmented, y
+
+
+
+# load test data
+X_test = np.empty((len(partition['test']), 224, 224, 3), dtype=np.float16)
+y_test = np.empty((len(partition['test']) - len(partition['test']) % batch_size, 14), dtype=np.float16)
+
+for i, npy in enumerate(partition['test']):
+    if (i < len(y_test)):
+        # round to batch_size
+        y_test[i, :] = labels[npy]
+
+print("len of result is", len(y_test))
+y_pred_list = np.empty((len(models_file_name), len(partition['test']), 14), dtype=np.float16)
+
+# individual models
+for index, current_model_file in enumerate(models_file_name):
+    print(current_model_file)
+#    model = load_model(current_model_file)
+    model = azure_chestxray_keras_utils.build_model(keras_contrib.applications.densenet.DenseNetImageNet121); model.load_weights(current_model_file)
+    print('evaluation for model', current_model_file)
+    # y_pred = model.predict(X_test)
+
+    y_pred = model.predict_generator(generator=DataGenSequence(labels, partition['test'], current_state='test'),
+                                     workers=32, verbose=1, max_queue_size=1)
+    print("result shape", y_pred.shape)
+
+    # add one fake row of ones in both test and pred values to avoid:
+    # ValueError: Only one class present in y_true. ROC AUC score is not defined in that case.
+    y_test = np.insert(y_test, 0, np.ones((y_test.shape[1],)), 0)
+    y_pred = np.insert(y_pred, 0, np.ones((y_pred.shape[1],)), 0)
+
+    df = pd.DataFrame(columns=['Disease', 'Our AUC Score', 'Stanford AUC Score'])
+    for d in range(14):
+        df.loc[d] = [pathologies_name_list[d],
+                     metrics.roc_auc_score(y_test[:, d], y_pred[:, d]),
+                     stanford_result[d]]
+
+    df['Delta'] = df['Stanford AUC Score'] - df['Our AUC Score']
+    df.to_csv(current_model_file + ".csv", index=False)
+    print(df)
\ No newline at end of file
diff --git a/Predict-Lung-Disease-master/040_cam_simple.py b/Predict-Lung-Disease-master/040_cam_simple.py
new file mode 100644
index 0000000..d58cc91
--- /dev/null
+++ b/Predict-Lung-Disease-master/040_cam_simple.py
@@ -0,0 +1,74 @@
+import sys, os
+import cv2
+import matplotlib
+import keras_contrib
+import azure_chestxray_cam, azure_chestxray_utils, azure_chestxray_keras_utils
+import keras_contrib
+from keras.models import Model
+
+
+path = os.getcwd()+r'\azure-share'
+amlWBSharedDir = path
+
+prj_consts = azure_chestxray_utils.chestxray_consts()
+
+data_base_output_dir=os.path.join(amlWBSharedDir,
+                                  os.path.join(*(prj_consts.BASE_OUTPUT_DIR_list)))
+data_base_input_dir=os.path.join(amlWBSharedDir,
+                                 os.path.join(*(prj_consts.BASE_INPUT_DIR_list)))
+
+# "quality" models, fully trained on all training data
+fully_trained_weights_dir=os.path.join(data_base_output_dir,
+    os.path.join(*(prj_consts.FULLY_PRETRAINED_MODEL_DIR_list)))
+
+test_images_dir=os.path.join(data_base_input_dir,
+    os.path.join(*(['test_images'])))
+
+test_images=azure_chestxray_utils.get_files_in_dir(test_images_dir)
+
+nih_chest_xray_data_dir=os.path.join(data_base_input_dir,
+                                     os.path.join(*(prj_consts.ChestXray_IMAGES_DIR_list)))
+
+chestXray_images=azure_chestxray_utils.get_files_in_dir(nih_chest_xray_data_dir)
+
+
+model = azure_chestxray_keras_utils.build_model(keras_contrib.applications.densenet.DenseNetImageNet121)
+
+model_file_name = prj_consts.PRETRAINED_DENSENET201_IMAGENET_CHESTXRAY_MODEL_FILE_NAME
+model_file_name = 'weights_only_azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5'
+model.load_weights(os.path.join(fully_trained_weights_dir, model_file_name))
+
+
+cv2_image = cv2.imread(os.path.join(test_images_dir,test_images[3]))
+
+
+
+predictions = model.predict(cv2_image[None,:,:,:])
+print(predictions)
+conv_map_model = Model(inputs=model.input, outputs=model.get_layer(index=-3).output)
+conv_features = conv_map_model.predict(cv2_image[None,:,:,:])
+conv_features = conv_features[0, :, :, :] #np.squeeze(conv_features)
+class_weights = model.layers[-1].get_weights()
+
+cv2_image = cv2.imread(os.path.join(test_images_dir,test_images[3]))
+
+azure_chestxray_utils.print_image_stats_by_channel(cv2_image)
+cv2_image = azure_chestxray_utils.normalize_nd_array(cv2_image)
+cv2_image = 255*cv2_image
+cv2_image=cv2_image.astype('uint8')
+azure_chestxray_utils.print_image_stats_by_channel(cv2_image)
+
+predictions, cam_image, predicted_disease_index = \
+azure_chestxray_cam.get_score_and_cam_picture(cv2_image, model)
+print(predictions)
+
+prj_consts.DISEASE_list[predicted_disease_index]
+print('likely disease: ', prj_consts.DISEASE_list[predicted_disease_index])
+print('likely disease prob ratio: ', \
+          predictions[predicted_disease_index]/sum(predictions))
+
+
+
+NIH_annotated_nodules = ['00000706_000.png', '00000702_000.png']
+azure_chestxray_cam.process_nih_data(NIH_annotated_nodules,
+                                   nih_chest_xray_data_dir, model)
\ No newline at end of file
diff --git a/Predict-Lung-Disease-master/README.md b/Predict-Lung-Disease-master/README.md
new file mode 100644
index 0000000..39bddb5
--- /dev/null
+++ b/Predict-Lung-Disease-master/README.md
@@ -0,0 +1,74 @@
+# Predict-Lung-Disease-through-Chest-X-Ray
+We obtain this repository by refactoring the [code](https://github.com/Azure/AzureChestXRay) for the blog post [Using Microsoft AI to Build a Lung-Disease Prediction Model using Chest X-Ray Images](https://blogs.technet.microsoft.com/machinelearning/2018/03/07/using-microsoft-ai-to-build-a-lung-disease-prediction-model-using-chest-x-ray-images/). This instruction aims to help newcomers build the system in a very short time.   
+# Installation
+1. Clone this repository
+   ```Shell
+   git clone https://github.com/svishwa/crowdcount-mcnn.git
+   ```
+   We'll call the directory that you cloned PredictLungDisease `ROOT`  
+  
+2. All essential dependencies should be installed：pickle, random, re, tqdm, cv2, numpy, pandas, sklearn, keras, tensorflow, keras_contrib, collections.counter.
+   
+# Data set up
+1. Download the NIH Chest X-ray Dataset from here:  
+   https://nihcc.app.box.com/v/ChestXray-NIHCC.  
+   You need to get all the image files (all the files under `images` folder in NIH Dataset), `Data_Entry_2017.csv` file, as well as the      Bounding Box data `BBox_List_2017.csv`.  
+
+2. Create Directory 
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC_other
+   ```  
+3. Save all images under `ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC`  
+
+4. Save `Data_Entry_2017.csv` and `BBox_List_2017.csv` under `ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC_other`  
+
+5. Process the Data
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/output/data_partitions
+   ```  
+   Run `000_preprocess.py` to create `*.pickle` files under this directory 
+# Test  
+1. We have provided the pretrained-model `azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5` under `ROOT/azure-share/chestxray/output/fully_trained_models`. You can also download it separately from [here](https://chestxray.blob.core.windows.net/chestxraytutorial/tutorial_xray/chexray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5).  
+
+2. Run `020_evaluate.py` and it will create `weights_only_azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5` saving weights of the pretrained-model under the same directory.
+
+3. Below is the result showing the AUC score of all the 14 diseases:  
+
+   | Disease            | Our AUC Score    | Stanford AUC Score | Delta     
+   |--------------------|------------------|--------------------|-----------:
+   | Atelectasis        | 0.822334         | 0.8094             | -0.012934 
+   | Cardiomegaly       | 0.933610         | 0.9248             | -0.008810 
+   | Effusion           | 0.882471         | 0.8638             | -0.018671 
+   | Infiltration       | 0.744504         | 0.7345             | -0.010004 
+   | Mass               | 0.858467         | 0.8676             |  0.009133 
+   | Nodule             | 0.784230         | 0.7802             | -0.004030 
+   | Pneumonia          | 0.800054         | 0.7680             | -0.032054 
+   | Pneumothorax       | 0.829764         | 0.8887             |  0.058936 
+   | Consolidation      | 0.811969         | 0.7901             | -0.021869 
+   | Edema              | 0.894102         | 0.8878             | -0.006302 
+   | Emphysema          | 0.847477         | 0.9371             |  0.089623
+   | Fibrosis           | 0.882602         | 0.8047             | -0.077902 
+   | Pleural Thickening | 1.000000         | 0.8062             | -0.193800 
+   | Hernia             | 0.916610         | 0.9164             | -0.000210   
+   
+# Visualization    
+1. Create Folder Test
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/test_images
+   ```  
+   Copy any number of images under `ChestXray-NIHCC` to `test_images` and resize them to 224x224 pixels.  
+
+2. Run `004_cam_simple.py` and it will output a Class Activation Map(CAM). The CAM lets us see which regions in the image were relevant to this class.  
+
+   ![这里随便写文字](https://github.com/fatLime/Predict-Lung-Disease/blob/master/image.png)
+
+# Referenced Paper  
+- Baseline result: https://arxiv.org/abs/1705.02315
+- Image Localization: http://arxiv.org/abs/1512.04150   
+- The original chexnet paper mentioned in [StanfordML website](https://stanfordmlgroup.github.io/projects/chexnet/) as well as their [paper](https://arxiv.org/abs/1711.05225).
+- http://cs231n.stanford.edu/reports/2017/pdfs/527.pdf for pre-processing the data
+- https://arxiv.org/abs/1711.08760 for some other thoughts on the model architecture and the relationship between different diseases
+ 
+# Notes
+  Please contact yanhaotian@bupt.edu.cn if you have any problem.
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png
new file mode 100644
index 0000000..b541ee2
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png differ
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png
new file mode 100644
index 0000000..10ebc70
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png differ
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle
new file mode 100644
index 0000000..cdfa519
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle
new file mode 100644
index 0000000..e7d2a81
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle
new file mode 100644
index 0000000..dc450cf
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5 b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5
new file mode 100644
index 0000000..ac6f4bf
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5 differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png b/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png
new file mode 100644
index 0000000..b541ee2
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000702_000.png differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png b/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png
new file mode 100644
index 0000000..10ebc70
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/data/ChestX-ray8/test_images/00000706_000.png differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle
new file mode 100644
index 0000000..cdfa519
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/labels14_unormalized_cleaned.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle
new file mode 100644
index 0000000..e7d2a81
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/partition14_unormalized_cleaned.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle
new file mode 100644
index 0000000..dc450cf
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/output/data_partitions/train_test_valid_data_partitions.pickle differ
diff --git a/Predict-Lung-Disease-master/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5 b/Predict-Lung-Disease-master/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5
new file mode 100644
index 0000000..ac6f4bf
Binary files /dev/null and b/Predict-Lung-Disease-master/azure-share/chestxray/output/fully_trained_models/azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5 differ
diff --git a/Predict-Lung-Disease-master/image.png b/Predict-Lung-Disease-master/image.png
new file mode 100644
index 0000000..12daa3d
Binary files /dev/null and b/Predict-Lung-Disease-master/image.png differ
diff --git a/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_cam.cpython-37.pyc b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_cam.cpython-37.pyc
new file mode 100644
index 0000000..6cb4f0b
Binary files /dev/null and b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_cam.cpython-37.pyc differ
diff --git a/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_keras_utils.cpython-37.pyc b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_keras_utils.cpython-37.pyc
new file mode 100644
index 0000000..6a6392c
Binary files /dev/null and b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_keras_utils.cpython-37.pyc differ
diff --git a/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_utils.cpython-37.pyc b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_utils.cpython-37.pyc
new file mode 100644
index 0000000..de31085
Binary files /dev/null and b/Predict-Lung-Disease-master/src/__pycache__/azure_chestxray_utils.cpython-37.pyc differ
diff --git a/Predict-Lung-Disease-master/src/azure_chestxray_cam.py b/Predict-Lung-Disease-master/src/azure_chestxray_cam.py
new file mode 100644
index 0000000..2235b1c
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/azure_chestxray_cam.py
@@ -0,0 +1,161 @@
+### Copyright (C) Microsoft Corporation.  
+
+import keras.backend as K
+import sys, os, io
+import numpy as np
+import cv2
+
+import matplotlib
+matplotlib.use('agg')
+
+paths_to_append = [os.path.join(os.getcwd(), os.path.join(*(['Code',  'src'])))]
+def add_path_to_sys_path(path_to_append):
+    if not (any(path_to_append in paths for paths in sys.path)):
+        sys.path.append(path_to_append)
+[add_path_to_sys_path(crt_path) for crt_path in paths_to_append]
+
+import azure_chestxray_utils
+
+
+def get_score_and_cam_picture(cv2_input_image, DenseNetImageNet121_model):
+# based on https://github.com/jacobgil/keras-cam/blob/master/cam.py
+    width, height, _ = cv2_input_image.shape
+    class_weights = DenseNetImageNet121_model.layers[-1].get_weights()[0]
+    final_conv_layer = DenseNetImageNet121_model.layers[-3]
+    get_output = K.function([DenseNetImageNet121_model.layers[0].input], 
+                            [final_conv_layer.output, \
+                             DenseNetImageNet121_model.layers[-1].output])
+    [conv_outputs, prediction] = get_output([cv2_input_image[None,:,:,:]])
+    conv_outputs = conv_outputs[0, :, :, :]
+    prediction = prediction[0,:]
+    
+    #Create the class activation map.
+    predicted_disease = np.argmax(prediction)
+    cam = np.zeros(dtype = np.float32, shape = conv_outputs.shape[:2])
+    for i, w in enumerate(class_weights[:, predicted_disease]):
+            cam += w * conv_outputs[:, :, i]
+    
+    return prediction, cam, predicted_disease
+
+
+def process_cam_image(crt_cam_image, xray_image, crt_alpha = .5):
+    im_width, im_height, _ = xray_image.shape
+    crt_cam_image = cv2.resize(crt_cam_image, (im_width, im_height),
+                               interpolation=cv2.INTER_CUBIC)
+    
+#     do some gamma enhancement, e is too much
+    crt_cam_image = np.power(1.1, crt_cam_image)
+    crt_cam_image = azure_chestxray_utils.normalize_nd_array(crt_cam_image)
+    # crt_cam_image[np.where(crt_cam_image < 0.5)] = 0 
+    crt_cam_image = 255*crt_cam_image
+
+    # make cam an rgb image
+    empty_image_channel = np.zeros(dtype = np.float32, shape = crt_cam_image.shape[:2])
+    crt_cam_image = cv2.merge((crt_cam_image,empty_image_channel,empty_image_channel))
+    
+    blended_image = cv2.addWeighted(xray_image.astype('uint8'),crt_alpha,
+                                    crt_cam_image.astype('uint8'),(1-crt_alpha),0)
+    return(blended_image)
+
+def plot_cam_results(crt_blended_image, crt_cam_image, crt_xray_image, map_caption):
+    import matplotlib
+    matplotlib.use('TkAgg')
+    import matplotlib.pyplot as plt
+    fig = plt.figure(figsize = (15,7))
+
+    ax1 = fig.add_subplot(2, 3, 1)
+    ax1.imshow(crt_xray_image, cmap = 'gray', interpolation = 'bicubic')
+    ax1.set_title('Orig X Ray')
+    plt.axis('off')
+
+
+    ax2 = fig.add_subplot(2,3, 2)
+    cam_plot = ax2.imshow(crt_cam_image, cmap=plt.get_cmap('OrRd'), interpolation = 'bicubic')
+    plt.colorbar(cam_plot, ax=ax2)
+    ax2.set_title('Activation Map')
+    plt.axis('off')
+
+
+    ax3 = fig.add_subplot(2,3, 3)
+    blended_plot = ax3.imshow(crt_blended_image, interpolation = 'bicubic')
+    plt.colorbar(cam_plot, ax=ax3)
+    ax3.set_title(map_caption)
+    plt.axis('off')
+
+    plt.show()
+
+    # serialize blended image plot padded in the x/y-direction
+    image_as_BytesIO = io.BytesIO()
+    x_direction_pad = 1.05;y_direction_pad=1.2
+    extent = ax3.get_window_extent().transformed(fig.dpi_scale_trans.inverted())
+    fig.savefig(image_as_BytesIO, 
+                bbox_inches=extent.expanded(x_direction_pad, 
+                                            y_direction_pad),
+               format='png')
+    image_as_BytesIO.seek(0)
+    return(image_as_BytesIO)
+    
+
+    
+def process_xray_image(crt_xray_image, DenseNetImageNet121_model):
+
+#     print(crt_xray_image.shape)
+    crt_xray_image = azure_chestxray_utils.normalize_nd_array(crt_xray_image)
+    crt_xray_image = 255*crt_xray_image
+    crt_xray_image=crt_xray_image.astype('uint8')
+
+    crt_predictions, crt_cam_image, predicted_disease_index =get_score_and_cam_picture(crt_xray_image, DenseNetImageNet121_model)
+    
+    prj_consts = azure_chestxray_utils.chestxray_consts()
+    likely_disease=prj_consts.DISEASE_list[predicted_disease_index]
+    likely_disease_prob = 100*crt_predictions[predicted_disease_index]
+    likely_disease_prob_ratio=100*crt_predictions[predicted_disease_index]/sum(crt_predictions)
+    print('predictions: ', crt_predictions)
+    print('likely disease: ', likely_disease)
+    print('likely disease prob: ', likely_disease_prob)
+    print('likely disease prob ratio: ', likely_disease_prob_ratio)
+    
+    crt_blended_image = process_cam_image(crt_cam_image, crt_xray_image)
+    plot_cam_results(crt_blended_image, crt_cam_image, crt_xray_image,
+                    str(likely_disease)+ ' ' +
+                    "{0:.1f}".format(likely_disease_prob)+ '% (weight ' +
+                    "{0:.1f}".format(likely_disease_prob_ratio)+ '%)')
+
+def process_nih_data(nih_data_files, NIH_data_dir, DenseNetImageNet121_model):
+    for crt_image in nih_data_files:
+        # print(crt_image)
+        prj_consts = azure_chestxray_utils.chestxray_consts()
+
+        crt_xray_image = cv2.imread(os.path.join(NIH_data_dir,crt_image))
+        crt_xray_image = cv2.resize(crt_xray_image, 
+                                    (prj_consts.CHESTXRAY_MODEL_EXPECTED_IMAGE_HEIGHT, 
+                                     prj_consts.CHESTXRAY_MODEL_EXPECTED_IMAGE_WIDTH)).astype(np.float32)
+
+        process_xray_image(crt_xray_image, DenseNetImageNet121_model )   
+        
+if __name__=="__main__":
+    #FIXME
+    # add example/test code here
+
+
+
+    NIH_annotated_Cardiomegaly = ['00005066_030.png']
+    data_dir = ''
+    cv2_image = cv2.imread(os.path.join(data_dir,NIH_annotated_Cardiomegaly[0]))
+
+    azure_chestxray_utils.print_image_stats_by_channel(cv2_image)
+    cv2_image = azure_chestxray_utils.normalize_nd_array(cv2_image)
+    cv2_image = 255*cv2_image
+    cv2_image=cv2_image.astype('uint8')
+    azure_chestxray_utils.print_image_stats_by_channel(cv2_image)
+
+    predictions, cam_image, predicted_disease_index = get_score_and_cam_picture(cv2_image, model)
+    print(predictions)
+    prj_consts = azure_chestxray_utils.chestxray_consts()
+    print(prj_consts.DISEASE_list[predicted_disease_index])
+    print('likely disease: ', prj_consts.DISEASE_list[predicted_disease_index])
+    print('likely disease prob ratio: ',
+          predictions[predicted_disease_index]/sum(predictions))
+    blended_image = process_cam_image(cam_image, cv2_image)
+    plot_cam_results(blended_image, cam_image, cv2_image,
+                 prj_consts.DISEASE_list[predicted_disease_index])      
\ No newline at end of file
diff --git a/Predict-Lung-Disease-master/src/azure_chestxray_keras_utils.py b/Predict-Lung-Disease-master/src/azure_chestxray_keras_utils.py
new file mode 100644
index 0000000..a3c2e59
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/azure_chestxray_keras_utils.py
@@ -0,0 +1,29 @@
+### Copyright (C) Microsoft Corporation.  
+
+from keras.layers import Dense
+from keras.models import Model
+from keras_contrib.applications.densenet import DenseNetImageNet121
+import keras_contrib
+
+def build_model(crt_densenet_function):
+    """
+
+    Returns: a model with specified weights
+
+    """
+    # define the model, use pre-trained weights for image_net
+    base_model = crt_densenet_function(input_shape=(224, 224, 3),
+                                     weights='imagenet',
+                                     include_top=False,
+                                     pooling='avg')
+
+    x = base_model.output
+    predictions = Dense(14, activation='sigmoid')(x)
+    model = Model(inputs=base_model.input, outputs=predictions)
+    return model
+
+if __name__=="__main__":        
+    model = build_model(DenseNetImageNet121)
+    print(model.summary())    
+    model = build_model(keras_contrib.applications.densenet.DenseNetImageNet201)
+    print(model.summary())
\ No newline at end of file
diff --git a/Predict-Lung-Disease-master/src/azure_chestxray_pytorch_utils.py b/Predict-Lung-Disease-master/src/azure_chestxray_pytorch_utils.py
new file mode 100644
index 0000000..e69de29
diff --git a/Predict-Lung-Disease-master/src/azure_chestxray_utils.py b/Predict-Lung-Disease-master/src/azure_chestxray_utils.py
new file mode 100644
index 0000000..6587879
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/azure_chestxray_utils.py
@@ -0,0 +1,59 @@
+### Copyright (C) Microsoft Corporation.  
+
+import os
+import numpy as np
+
+class chestxray_consts(object):
+    DISEASE_list = ['Atelectasis', 'Cardiomegaly', 'Effusion', 'Infiltration', 'Mass', 'Nodule', 'Pneumonia',
+                'Pneumothorax',
+                'Consolidation', 'Edema', 'Emphysema', 'Fibrosis', 'Pleural Thickening', 'Hernia']
+    
+    PRETRAINED_DENSENET201_IMAGENET_CHESTXRAY_MODEL_FILE_NAME =  'chexnet_14_weights_multigpu_contribmodel_121layer_712split_epoch_011_val_loss_153.9783.hdf5'
+    FULLY_PRETRAINED_MODEL_DIR_list = [ 'fully_trained_models']
+
+
+    CHESTXRAY_MODEL_EXPECTED_IMAGE_HEIGHT  = 224
+    CHESTXRAY_MODEL_EXPECTED_IMAGE_WIDTH = 224
+
+    BASE_INPUT_DIR_list = ['chestxray', 'data', 'ChestX-ray8']
+    BASE_OUTPUT_DIR_list = ['chestxray', 'output']
+    CREDENTIALS_DIR_list = ['code', 'notShared']
+
+    SRC_DIR_list = ['Code',  'src']
+    ChestXray_IMAGES_DIR_list = ['ChestXray-NIHCC']
+    ChestXray_OTHER_DATA_DIR_list = ['ChestXray-NIHCC_other']
+    PROCESSED_IMAGES_DIR_list = ['processed_npy14']
+    DATA_PARTITIONS_DIR_list = ['data_partitions']
+    MODEL_WEIGHTS_DIR_list = ['weights_tmpdir']
+
+    def __setattr__(self, *_):
+        raise TypeError
+
+
+# os agnostic 'ls' function
+def get_files_in_dir(crt_dir):
+        return( [f for f in os.listdir(crt_dir) if os.path.isfile(os.path.join(crt_dir, f))])
+        
+       
+    
+def normalize_nd_array(crt_array):
+    # Normalised [0,1]
+    crt_array = crt_array - np.min(crt_array)
+    return(crt_array/np.ptp(crt_array))
+
+def print_image_stats_by_channel(crt_image):
+    print('min:')
+    print(np.amin(crt_image[:,:,0]), 
+          np.amin(crt_image[:,:,1]),
+          np.amin(crt_image[:,:,2]))
+    print('max:')
+    print(np.amax(crt_image[:,:,0]), 
+          np.amax(crt_image[:,:,1]),
+          np.amax(crt_image[:,:,2]))        
+
+        
+        
+if __name__=="__main__":        
+    prj_consts = chestxray_consts()
+    print('model_expected_image_height = ', prj_consts.CHESTXRAY_MODEL_EXPECTED_IMAGE_HEIGHT)
+    print('model_expected_image_width = ', prj_consts.CHESTXRAY_MODEL_EXPECTED_IMAGE_WIDTH)
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/blacklist_non_PA_AP_view.csv b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_non_PA_AP_view.csv
new file mode 100644
index 0000000..e986dcf
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_non_PA_AP_view.csv
@@ -0,0 +1,56 @@
+00000591_003.png
+00001136_001.png
+00001153_005.png
+00001602_000.png
+00001803_003.png
+00002097_000.png
+00002117_003.png
+00002354_000.png
+00002592_003.png
+00002639_009.png
+00003023_000.png
+00003094_002.png
+00004533_004.png
+00004808_001.png
+00004906_000.png
+00005192_001.png
+00005260_000.png
+00005286_001.png
+00006462_008.png
+00006836_008.png
+00006851_004.png
+00007113_001.png
+00007152_006.png
+00007160_002.png
+00007454_001.png
+00007482_010.png
+00007716_007.png
+00008016_000.png
+00008082_000.png
+00009198_002.png
+00009368_010.png
+00009368_011.png
+00009584_002.png
+00009889_038.png
+00010007_121.png
+00010065_000.png
+00012249_001.png
+00012388_002.png
+00012907_007.png
+00013160_000.png
+00013670_137.png
+00013714_001.png
+00014294_015.png
+00014675_034.png
+00014963_000.png
+00015054_000.png
+00015078_007.png
+00016233_004.png
+00016637_000.png
+00017753_022.png
+00017915_003.png
+00020373_002.png
+00020644_000.png
+00025381_004.png
+00026806_000.png
+00029476_000.png
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/blacklist_other_images_with_lower_quality.csv b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_other_images_with_lower_quality.csv
new file mode 100644
index 0000000..eaf990a
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_other_images_with_lower_quality.csv
@@ -0,0 +1,1094 @@
+00000032_013.png
+00000032_023.png
+00000032_024.png
+00000032_055.png
+00000032_058.png
+00000116_007.png
+00000244_000.png
+00000244_002.png
+00000248_013.png
+00000248_018.png
+00000248_019.png
+00000467_013.png
+00000468_060.png
+00000565_000.png
+00000583_005.png
+00000583_007.png
+00000583_009.png
+00000583_019.png
+00000583_024.png
+00000627_030.png
+00000627_036.png
+00000703_000.png
+00000831_008.png
+00000929_000.png
+00000929_001.png
+00000980_004.png
+00001029_003.png
+00001075_016.png
+00001075_020.png
+00001122_016.png
+00001122_017.png
+00001153_006.png
+00001157_002.png
+00001179_000.png
+00001179_001.png
+00001181_000.png
+00001223_000.png
+00001223_001.png
+00001249_004.png
+00001255_012.png
+00001255_035.png
+00001267_001.png
+00001278_009.png
+00001278_011.png
+00001437_038.png
+00001501_002.png
+00001564_000.png
+00001577_003.png
+00001595_000.png
+00001595_001.png
+00001595_002.png
+00001686_000.png
+00001686_001.png
+00001736_005.png
+00001736_007.png
+00001736_010.png
+00001736_014.png
+00001736_018.png
+00001736_021.png
+00001736_025.png
+00001736_026.png
+00001736_027.png
+00001787_002.png
+00001787_010.png
+00001814_004.png
+00001836_014.png
+00001855_000.png
+00001855_004.png
+00001855_009.png
+00001855_010.png
+00001855_011.png
+00001855_012.png
+00001855_014.png
+00001855_016.png
+00001855_018.png
+00001855_020.png
+00001855_021.png
+00001855_022.png
+00001855_023.png
+00001855_024.png
+00001855_025.png
+00001855_026.png
+00001855_027.png
+00001855_028.png
+00001855_029.png
+00001855_030.png
+00001855_032.png
+00001855_033.png
+00001855_034.png
+00001855_035.png
+00001855_037.png
+00001952_000.png
+00001952_001.png
+00001952_002.png
+00001952_007.png
+00001952_008.png
+00001986_010.png
+00002072_003.png
+00002072_004.png
+00002072_009.png
+00002072_010.png
+00002072_011.png
+00002072_014.png
+00002072_015.png
+00002072_018.png
+00002072_019.png
+00002208_001.png
+00002359_018.png
+00002366_001.png
+00002366_002.png
+00002437_036.png
+00002437_037.png
+00002529_007.png
+00002529_023.png
+00002529_025.png
+00002529_030.png
+00002545_001.png
+00002582_007.png
+00002594_001.png
+00002633_023.png
+00002636_000.png
+00002659_003.png
+00002675_005.png
+00002733_000.png
+00002763_023.png
+00002763_024.png
+00002892_004.png
+00002896_000.png
+00003004_000.png
+00003005_005.png
+00003029_018.png
+00003059_000.png
+00003060_000.png
+00003094_000.png
+00003094_003.png
+00003094_004.png
+00003094_005.png
+00003158_001.png
+00003186_003.png
+00003369_001.png
+00003465_000.png
+00003465_001.png
+00003465_002.png
+00003465_003.png
+00003465_004.png
+00003465_005.png
+00003465_006.png
+00003465_007.png
+00003465_008.png
+00003523_036.png
+00004276_000.png
+00004285_000.png
+00004309_006.png
+00004344_025.png
+00004360_020.png
+00004360_023.png
+00004472_000.png
+00004545_000.png
+00004660_000.png
+00004672_001.png
+00004703_000.png
+00004706_001.png
+00004792_000.png
+00004808_014.png
+00004808_094.png
+00004811_000.png
+00004928_006.png
+00005051_000.png
+00005094_009.png
+00005201_001.png
+00005204_001.png
+00005220_012.png
+00005220_015.png
+00005254_003.png
+00005254_004.png
+00005254_008.png
+00005298_013.png
+00005360_002.png
+00005573_004.png
+00005699_005.png
+00005712_008.png
+00005746_008.png
+00005748_000.png
+00005750_015.png
+00005750_016.png
+00005750_017.png
+00005877_000.png
+00005937_000.png
+00005975_001.png
+00006008_015.png
+00006015_000.png
+00006015_003.png
+00006039_022.png
+00006054_001.png
+00006127_000.png
+00006209_001.png
+00006220_002.png
+00006220_003.png
+00006220_004.png
+00006220_005.png
+00006220_006.png
+00006220_009.png
+00006271_002.png
+00006271_078.png
+00006271_093.png
+00006294_004.png
+00006296_011.png
+00006381_009.png
+00006391_001.png
+00006415_000.png
+00006446_012.png
+00006585_007.png
+00006754_008.png
+00006838_000.png
+00006850_019.png
+00006870_000.png
+00006904_007.png
+00006906_029.png
+00006906_031.png
+00006906_032.png
+00006917_000.png
+00006960_022.png
+00007001_001.png
+00007018_034.png
+00007018_035.png
+00007108_006.png
+00007152_008.png
+00007217_005.png
+00007269_000.png
+00007276_001.png
+00007276_002.png
+00007322_003.png
+00007322_005.png
+00007322_009.png
+00007322_020.png
+00007371_000.png
+00007438_000.png
+00007500_000.png
+00007500_001.png
+00007545_000.png
+00007558_004.png
+00007558_007.png
+00007624_036.png
+00007830_000.png
+00007830_001.png
+00007830_004.png
+00007830_005.png
+00007830_007.png
+00007830_010.png
+00007973_000.png
+00007985_000.png
+00008051_039.png
+00008051_050.png
+00008051_051.png
+00008295_010.png
+00008297_008.png
+00008297_013.png
+00008297_016.png
+00008314_000.png
+00008463_001.png
+00008522_057.png
+00008549_000.png
+00008640_000.png
+00008701_008.png
+00008911_006.png
+00008993_000.png
+00009218_020.png
+00009218_022.png
+00009282_000.png
+00009465_004.png
+00009508_004.png
+00009551_008.png
+00009551_022.png
+00009573_000.png
+00009608_045.png
+00009613_005.png
+00009621_000.png
+00009621_001.png
+00009621_002.png
+00009621_003.png
+00009621_004.png
+00009621_005.png
+00009621_006.png
+00009621_007.png
+00009702_006.png
+00009727_012.png
+00009727_013.png
+00009727_014.png
+00009727_018.png
+00009727_019.png
+00009727_020.png
+00009727_022.png
+00009727_023.png
+00009727_027.png
+00009727_028.png
+00009876_002.png
+00009886_000.png
+00009892_007.png
+00009892_046.png
+00009911_004.png
+00009953_016.png
+00010007_053.png
+00010007_060.png
+00010007_071.png
+00010007_074.png
+00010007_082.png
+00010007_103.png
+00010012_018.png
+00010012_026.png
+00010092_007.png
+00010092_043.png
+00010124_000.png
+00010294_007.png
+00010352_054.png
+00010352_074.png
+00010360_004.png
+00010384_005.png
+00010405_000.png
+00010405_001.png
+00010415_000.png
+00010435_002.png
+00010544_016.png
+00010544_027.png
+00010544_030.png
+00010693_027.png
+00010698_001.png
+00010698_013.png
+00010761_000.png
+00010773_014.png
+00010773_025.png
+00010790_039.png
+00010790_043.png
+00010790_045.png
+00010792_004.png
+00010805_002.png
+00010805_003.png
+00010805_004.png
+00010805_005.png
+00010805_006.png
+00010805_008.png
+00010805_009.png
+00010805_010.png
+00010805_011.png
+00010805_013.png
+00010805_015.png
+00010805_016.png
+00010805_017.png
+00010805_018.png
+00010805_019.png
+00010805_020.png
+00010805_023.png
+00010805_025.png
+00010805_037.png
+00010805_038.png
+00010805_040.png
+00010805_043.png
+00010805_045.png
+00010805_046.png
+00010805_047.png
+00010805_048.png
+00010805_050.png
+00010828_017.png
+00010843_000.png
+00010887_027.png
+00010960_001.png
+00010960_002.png
+00010995_006.png
+00010995_008.png
+00011007_000.png
+00011021_012.png
+00011064_000.png
+00011164_007.png
+00011237_095.png
+00011237_108.png
+00011379_002.png
+00011379_003.png
+00011379_004.png
+00011379_005.png
+00011379_006.png
+00011379_013.png
+00011379_018.png
+00011379_019.png
+00011379_022.png
+00011379_039.png
+00011379_041.png
+00011379_043.png
+00011379_045.png
+00011379_046.png
+00011379_047.png
+00011386_000.png
+00011391_016.png
+00011391_031.png
+00011391_032.png
+00011391_039.png
+00011391_041.png
+00011391_043.png
+00011391_047.png
+00011391_055.png
+00011436_009.png
+00011461_002.png
+00011553_002.png
+00011553_003.png
+00011553_004.png
+00011553_005.png
+00011553_006.png
+00011553_007.png
+00011553_009.png
+00011553_010.png
+00011553_011.png
+00011553_012.png
+00011553_013.png
+00011553_014.png
+00011553_015.png
+00011553_016.png
+00011553_017.png
+00011553_018.png
+00011553_019.png
+00011553_020.png
+00011553_022.png
+00011553_023.png
+00011553_024.png
+00011553_025.png
+00011553_026.png
+00011553_027.png
+00011553_028.png
+00011553_029.png
+00011553_030.png
+00011553_031.png
+00011553_032.png
+00011553_033.png
+00011553_034.png
+00011553_035.png
+00011553_036.png
+00011553_037.png
+00011553_038.png
+00011553_040.png
+00011553_041.png
+00011553_046.png
+00011553_047.png
+00011673_000.png
+00011677_001.png
+00011677_002.png
+00011702_024.png
+00011702_062.png
+00011731_003.png
+00011769_000.png
+00011925_047.png
+00011925_049.png
+00011925_051.png
+00011925_053.png
+00011925_055.png
+00011925_068.png
+00011925_071.png
+00011925_078.png
+00011947_000.png
+00011985_008.png
+00012141_013.png
+00012159_002.png
+00012162_001.png
+00012276_007.png
+00012276_009.png
+00012276_010.png
+00012276_013.png
+00012276_017.png
+00012276_018.png
+00012368_002.png
+00012470_011.png
+00012470_012.png
+00012515_002.png
+00012591_000.png
+00012605_000.png
+00012605_001.png
+00012628_017.png
+00012628_060.png
+00012648_001.png
+00012654_001.png
+00012662_000.png
+00012742_000.png
+00012742_001.png
+00012742_002.png
+00012798_000.png
+00012834_005.png
+00012834_007.png
+00012834_010.png
+00012834_085.png
+00012834_120.png
+00012834_137.png
+00012863_027.png
+00012863_039.png
+00013049_006.png
+00013049_007.png
+00013123_004.png
+00013152_004.png
+00013158_004.png
+00013249_004.png
+00013249_006.png
+00013249_007.png
+00013249_008.png
+00013249_013.png
+00013249_014.png
+00013249_017.png
+00013249_018.png
+00013249_028.png
+00013249_033.png
+00013249_036.png
+00013249_038.png
+00013249_041.png
+00013249_046.png
+00013401_000.png
+00013440_000.png
+00013499_004.png
+00013568_000.png
+00013601_013.png
+00013608_000.png
+00013608_002.png
+00013608_004.png
+00013608_016.png
+00013615_015.png
+00013615_025.png
+00013615_027.png
+00013615_049.png
+00013615_057.png
+00013615_060.png
+00013625_033.png
+00013641_014.png
+00013641_041.png
+00013670_146.png
+00013670_162.png
+00013670_163.png
+00013670_166.png
+00013670_167.png
+00013685_047.png
+00013774_027.png
+00013774_041.png
+00013774_042.png
+00013774_048.png
+00013894_010.png
+00013894_024.png
+00013894_025.png
+00013894_027.png
+00013896_004.png
+00013922_020.png
+00013922_021.png
+00013966_007.png
+00013993_016.png
+00013993_049.png
+00013993_099.png
+00014004_018.png
+00014004_023.png
+00014014_002.png
+00014014_005.png
+00014080_001.png
+00014112_019.png
+00014128_023.png
+00014192_000.png
+00014203_016.png
+00014203_026.png
+00014203_028.png
+00014203_029.png
+00014203_042.png
+00014203_044.png
+00014223_012.png
+00014245_001.png
+00014245_003.png
+00014314_001.png
+00014320_040.png
+00014320_043.png
+00014323_001.png
+00014323_002.png
+00014323_003.png
+00014332_004.png
+00014351_000.png
+00014352_001.png
+00014465_016.png
+00014474_002.png
+00014486_004.png
+00014509_000.png
+00014958_009.png
+00014982_000.png
+00015007_002.png
+00015007_003.png
+00015007_005.png
+00015007_006.png
+00015007_007.png
+00015007_008.png
+00015007_011.png
+00015024_003.png
+00015031_006.png
+00015031_022.png
+00015041_003.png
+00015112_004.png
+00015126_000.png
+00015151_001.png
+00015193_014.png
+00015213_000.png
+00015290_000.png
+00015391_001.png
+00015462_001.png
+00015462_002.png
+00015482_000.png
+00015530_071.png
+00015530_142.png
+00015564_011.png
+00015605_038.png
+00015605_051.png
+00015605_053.png
+00015605_055.png
+00015606_013.png
+00015606_050.png
+00015696_001.png
+00015758_000.png
+00015826_019.png
+00015923_000.png
+00015934_000.png
+00015986_000.png
+00015996_001.png
+00016009_046.png
+00016034_003.png
+00016051_003.png
+00016051_004.png
+00016133_000.png
+00016175_003.png
+00016175_006.png
+00016175_008.png
+00016184_027.png
+00016238_006.png
+00016292_000.png
+00016292_001.png
+00016292_002.png
+00016292_003.png
+00016292_004.png
+00016378_001.png
+00016410_006.png
+00016410_008.png
+00016410_055.png
+00016484_001.png
+00016484_005.png
+00016484_009.png
+00016484_011.png
+00016484_026.png
+00016522_019.png
+00016529_000.png
+00016638_003.png
+00016638_004.png
+00016653_000.png
+00016732_035.png
+00016784_002.png
+00016860_001.png
+00016860_005.png
+00016867_003.png
+00016918_005.png
+00017036_023.png
+00017110_012.png
+00017138_032.png
+00017207_002.png
+00017207_003.png
+00017207_008.png
+00017258_022.png
+00017258_023.png
+00017362_009.png
+00017392_000.png
+00017400_000.png
+00017403_007.png
+00017403_010.png
+00017424_034.png
+00017424_035.png
+00017424_036.png
+00017424_038.png
+00017424_041.png
+00017425_002.png
+00017425_006.png
+00017477_000.png
+00017504_024.png
+00017504_068.png
+00017538_001.png
+00017538_002.png
+00017541_025.png
+00017553_000.png
+00017561_001.png
+00017605_014.png
+00017606_020.png
+00017618_013.png
+00017625_000.png
+00017625_004.png
+00017641_004.png
+00017645_013.png
+00017648_000.png
+00017695_000.png
+00017753_026.png
+00017817_001.png
+00017817_002.png
+00017927_001.png
+00017941_005.png
+00017972_006.png
+00017972_014.png
+00017979_000.png
+00017999_000.png
+00018011_015.png
+00018044_020.png
+00018044_036.png
+00018044_040.png
+00018044_043.png
+00018069_000.png
+00018069_001.png
+00018091_012.png
+00018103_002.png
+00018103_007.png
+00018103_009.png
+00018104_004.png
+00018116_000.png
+00018121_000.png
+00018125_009.png
+00018126_024.png
+00018175_002.png
+00018191_000.png
+00018191_001.png
+00018213_001.png
+00018240_000.png
+00018251_001.png
+00018251_002.png
+00018251_003.png
+00018251_004.png
+00018251_005.png
+00018251_006.png
+00018251_007.png
+00018251_008.png
+00018251_009.png
+00018251_010.png
+00018251_011.png
+00018251_012.png
+00018251_013.png
+00018251_014.png
+00018253_089.png
+00018336_000.png
+00018437_001.png
+00018437_002.png
+00018445_002.png
+00018458_000.png
+00018486_000.png
+00018571_000.png
+00018573_000.png
+00018598_004.png
+00018610_002.png
+00018610_004.png
+00018614_001.png
+00018615_001.png
+00018778_001.png
+00018778_002.png
+00018778_005.png
+00018921_026.png
+00018921_027.png
+00018927_000.png
+00018949_001.png
+00019020_000.png
+00019045_000.png
+00019107_001.png
+00019124_011.png
+00019124_012.png
+00019150_007.png
+00019301_000.png
+00019390_002.png
+00019390_004.png
+00019534_000.png
+00019576_024.png
+00019576_063.png
+00019576_064.png
+00019576_065.png
+00019587_000.png
+00019592_010.png
+00019660_001.png
+00019707_010.png
+00019888_001.png
+00019928_000.png
+00019967_001.png
+00019967_002.png
+00019967_003.png
+00019967_004.png
+00019967_007.png
+00019967_008.png
+00019967_009.png
+00019967_011.png
+00019967_012.png
+00019967_013.png
+00019967_014.png
+00019967_017.png
+00019967_019.png
+00019967_020.png
+00019967_032.png
+00020006_001.png
+00020108_001.png
+00020110_000.png
+00020146_002.png
+00020213_011.png
+00020213_018.png
+00020213_060.png
+00020213_061.png
+00020213_113.png
+00020219_000.png
+00020326_013.png
+00020326_058.png
+00020348_000.png
+00020364_002.png
+00020364_003.png
+00020398_010.png
+00020438_007.png
+00020622_002.png
+00020631_009.png
+00020928_004.png
+00020928_014.png
+00020928_015.png
+00020945_022.png
+00021023_014.png
+00021044_000.png
+00021108_000.png
+00021201_042.png
+00021420_013.png
+00021420_028.png
+00021481_012.png
+00021506_001.png
+00021508_002.png
+00021510_000.png
+00021572_010.png
+00021695_003.png
+00021700_006.png
+00021729_000.png
+00021770_012.png
+00021770_014.png
+00021770_015.png
+00021770_016.png
+00021811_003.png
+00021835_029.png
+00021901_005.png
+00021917_000.png
+00021942_006.png
+00021990_002.png
+00022010_001.png
+00022051_000.png
+00022174_000.png
+00022174_001.png
+00022245_011.png
+00022283_029.png
+00022339_000.png
+00022416_052.png
+00022470_007.png
+00022486_000.png
+00022523_004.png
+00022523_005.png
+00022524_000.png
+00022528_007.png
+00022566_022.png
+00022599_004.png
+00022714_000.png
+00022723_000.png
+00022725_003.png
+00022727_001.png
+00022815_004.png
+00022815_015.png
+00022815_020.png
+00022815_031.png
+00022815_037.png
+00022815_058.png
+00022815_068.png
+00022815_073.png
+00022815_079.png
+00022872_001.png
+00022872_002.png
+00022975_004.png
+00023027_000.png
+00023068_015.png
+00023129_000.png
+00023160_003.png
+00023176_019.png
+00023192_000.png
+00023195_000.png
+00023197_000.png
+00023254_003.png
+00023271_016.png
+00023325_037.png
+00023325_039.png
+00025066_000.png
+00025203_000.png
+00025223_000.png
+00025290_014.png
+00025445_001.png
+00025513_001.png
+00025513_005.png
+00025513_006.png
+00025513_007.png
+00025513_008.png
+00025513_009.png
+00025513_010.png
+00025513_011.png
+00025513_012.png
+00025513_013.png
+00025513_014.png
+00025529_010.png
+00025628_024.png
+00025628_026.png
+00025628_027.png
+00025664_037.png
+00025665_000.png
+00025691_000.png
+00025691_002.png
+00025697_001.png
+00025704_000.png
+00025796_000.png
+00025809_001.png
+00025839_010.png
+00025839_012.png
+00025932_001.png
+00025958_000.png
+00025958_002.png
+00025958_003.png
+00025958_006.png
+00026068_000.png
+00026068_001.png
+00026092_003.png
+00026098_028.png
+00026099_041.png
+00026114_000.png
+00026115_000.png
+00026159_000.png
+00026167_008.png
+00026194_001.png
+00026194_004.png
+00026194_007.png
+00026194_008.png
+00026194_009.png
+00026194_010.png
+00026194_011.png
+00026194_012.png
+00026194_014.png
+00026194_015.png
+00026194_018.png
+00026232_030.png
+00026262_000.png
+00026346_015.png
+00026349_005.png
+00026382_009.png
+00026431_000.png
+00026474_003.png
+00026506_000.png
+00026538_025.png
+00026621_000.png
+00026634_000.png
+00026666_000.png
+00026701_000.png
+00026758_000.png
+00026801_005.png
+00026818_020.png
+00026867_002.png
+00026867_004.png
+00026911_004.png
+00026925_006.png
+00026925_011.png
+00026925_015.png
+00026963_032.png
+00026971_026.png
+00026993_003.png
+00026993_004.png
+00027072_000.png
+00027196_009.png
+00027196_010.png
+00027213_001.png
+00027213_008.png
+00027213_009.png
+00027213_010.png
+00027213_076.png
+00027213_079.png
+00027299_006.png
+00027299_007.png
+00027415_009.png
+00027415_011.png
+00027415_028.png
+00027415_029.png
+00027415_037.png
+00027415_046.png
+00027415_047.png
+00027415_049.png
+00027415_059.png
+00027415_068.png
+00027415_069.png
+00027415_072.png
+00027415_073.png
+00027415_075.png
+00027415_077.png
+00027441_012.png
+00027441_017.png
+00027441_019.png
+00027441_024.png
+00027442_008.png
+00027464_024.png
+00027465_008.png
+00027524_000.png
+00027618_012.png
+00027623_006.png
+00027639_000.png
+00027639_001.png
+00027639_002.png
+00027639_003.png
+00027677_000.png
+00027710_000.png
+00027725_021.png
+00027725_035.png
+00027726_016.png
+00027726_019.png
+00027726_020.png
+00027726_021.png
+00027726_050.png
+00027726_051.png
+00027765_000.png
+00027765_002.png
+00027952_004.png
+00027981_000.png
+00027981_001.png
+00027981_002.png
+00028076_000.png
+00028092_000.png
+00028201_000.png
+00028211_012.png
+00028301_002.png
+00028341_001.png
+00028341_002.png
+00028341_003.png
+00028341_004.png
+00028341_005.png
+00028341_006.png
+00028341_007.png
+00028341_008.png
+00028341_009.png
+00028341_010.png
+00028341_011.png
+00028341_012.png
+00028389_000.png
+00028450_000.png
+00028454_011.png
+00028454_013.png
+00028474_000.png
+00028657_000.png
+00028799_000.png
+00028829_002.png
+00028873_017.png
+00028873_019.png
+00028873_020.png
+00028882_004.png
+00028961_006.png
+00028961_008.png
+00028996_002.png
+00028996_003.png
+00028996_004.png
+00029174_002.png
+00029222_003.png
+00029235_001.png
+00029245_002.png
+00029276_004.png
+00029404_000.png
+00029404_002.png
+00029404_003.png
+00029404_004.png
+00029404_005.png
+00029404_006.png
+00029404_007.png
+00029404_008.png
+00029404_010.png
+00029476_003.png
+00029596_012.png
+00029627_000.png
+00029813_029.png
+00029943_022.png
+00030079_020.png
+00030079_031.png
+00030206_011.png
+00030209_011.png
+00030213_000.png
+00030245_001.png
+00030320_004.png
+00030320_006.png
+00030323_038.png
+00030410_005.png
+00030412_002.png
+00030609_000.png
+00030609_001.png
+00030609_002.png
+00030609_003.png
+00030609_006.png
+00030609_008.png
+00030609_009.png
+00030609_010.png
+00030609_011.png
+00030609_017.png
+00030609_021.png
+00030609_023.png
+00030609_026.png
+00030786_004.png
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/blacklist_rotated_images.csv b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_rotated_images.csv
new file mode 100644
index 0000000..c8a4148
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/finding_lungs/blacklist_rotated_images.csv
@@ -0,0 +1,18 @@
+00001255_007.png
+00001814_001.png
+00002180_000.png
+00002815_003.png
+00003693_005.png
+00005823_000.png
+00007188_002.png
+00008051_036.png
+00008468_003.png
+00009889_023.png
+00009984_001.png
+00011460_066.png
+00013299_000.png
+00013431_000.png
+00017258_011.png
+00017606_037.png
+00019620_001.png
+00026701_001.png
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/finding_lungs_DL_approach.py b/Predict-Lung-Disease-master/src/finding_lungs/finding_lungs_DL_approach.py
new file mode 100644
index 0000000..314cfce
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/finding_lungs/finding_lungs_DL_approach.py
@@ -0,0 +1,166 @@
+import os
+
+os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"  # see issue #152
+os.environ["CUDA_VISIBLE_DEVICES"] = "0"
+import pandas as pd
+from keras.models import load_model
+from skimage import exposure, img_as_float
+from skimage import transform
+import numpy as np
+
+from skimage import measure
+import lungs_finder as lf
+import cv2
+
+# for lung detection
+left_edge = 0
+right_edge = 256
+top_edge = 0
+bottom_edge = 256
+margin = 12
+
+row_size = 256
+col_size = 256
+
+# Path to csv-file. File should contain X-ray filenames as first column,
+# mask filenames as second column.
+
+out_folder_matched_img = os.path.join("/mnt", "MyAzureFileShare", "Data", "ChestXRay", "images_centered")
+out_folder_mismatched_image = os.path.join("/mnt", "MyAzureFileShare", "Data", "ChestXRay",
+                                           "images_centered_mismatched_by_both")
+csv_path = os.path.join("/mnt", "MyAzureFileShare", "Data", "ChestXRay", "Data_Entry_2017.csv")
+# Path to the folder with images. Images will be read from path + path_from_csv
+img_path = os.path.join("/mnt", "MyAzureFileShare", "Data", "ChestXRay", "images")
+mis_detected_csv_path = os.path.join("/mnt", "MyAzureFileShare", "Data", "ChestXRay", "mis_detected.csv")
+df = pd.read_csv(csv_path)
+
+# Load test data
+im_shape = (256, 256)
+
+# Load model
+# plt.figure(figsize=(10, 10))
+model_name = './trained_model.hdf5.bak'
+UNet = load_model(model_name)
+
+threshold = 0.85
+# list to save the mis detected images
+image_misdetect_list = []
+
+
+def finding_lungs_non_DL_approach_and_save(image, file_name):
+    # print(row.columns.values)
+    # file_name = row[0]
+    # print("line is", file_name, image.shape)
+    # when reading from txt there is something in the end so we need to eliminate that
+    # image = cv2.imread(os.path.join("Z:\\", "Data", "ChestXRay", "images", file_name), 0)
+
+    img_height = image.shape[0]
+    img_width = image.shape[1]
+    # Get both lungs image. It uses HOG as main method,
+    # but if HOG found nothing it uses HAAR or LBP.
+    found_lungs = lf.get_lungs(image)
+
+    # this can be written in a more concise way but we just keep it a bit redundant for easy reading
+    if found_lungs is not None and found_lungs.shape[0] > img_height / 2 and found_lungs.shape[1] > img_width / 2:
+        # print(found_lungs.shape)
+        found_lungs_resized = cv2.resize(found_lungs, im_shape)
+        # cv2.imshow(file_name, found_lungs)
+        # code = cv2.waitKey(0)
+        cv2.imwrite(os.path.join(out_folder_matched_img, file_name), found_lungs_resized)
+        return True
+    else:
+        cv2.imwrite(os.path.join(out_folder_mismatched_image, file_name), cv2.resize(image, im_shape))
+        return False
+
+
+for index, item in df.iterrows():
+    # X, y = loadDataGeneral(current_df, path, im_shape)
+    raw_img = cv2.imread(os.path.join(img_path, item['Image Index']))
+
+    img = img_as_float(raw_img)[:, :, 0]
+    img = transform.resize(img, im_shape)
+    img = exposure.equalize_hist(img)
+    # img = np.expand_dims(img, -1)
+    img -= img.mean()
+    img /= img.std()
+
+    file_name = item['Image Index']
+    X = np.expand_dims(img, axis=0)
+    X = np.expand_dims(X, axis=-1)
+    n_test = X.shape[0]
+    inp_shape = X[0].shape
+
+    # img = exposure.rescale_intensity(np.squeeze(X), out_range=(0, 1))
+
+    # print("size of img is", img.shape)
+    prediction = UNet.predict(X)[..., 0].reshape(inp_shape[:2])
+
+    thresh_img = np.where(prediction > threshold, 1.0, 0.0)  # threshold the image
+
+    labels = measure.label(thresh_img)  # Different labels are displayed in different colors
+    label_vals = np.unique(labels)
+    # print(label_vals)
+    regions = measure.regionprops(labels)
+    good_labels = []
+    global_B_box = []
+    for prop in regions:
+        B = prop.bbox
+        if B[2] - B[0] > row_size / 4 and B[3] - B[1] > col_size / 6:  # make sure size of lung to avoid small areas
+            good_labels.append(prop.label)
+            global_B_box.append(B)
+
+    # print(len(good_labels))
+
+    DL_failed_detect_flag = False
+    if len(good_labels) == 2:
+
+        left_edge = np.clip(min(global_B_box[0][1] - margin, global_B_box[1][1] - margin), a_min=0, a_max=256)
+        right_edge = np.clip(max(global_B_box[0][3] + margin, global_B_box[1][3] + margin), a_min=0, a_max=256)
+        top_edge = np.clip(min(global_B_box[0][0] - margin, global_B_box[1][0] - margin), a_min=0, a_max=256)
+        bottom_edge = np.clip(max(global_B_box[0][2] + margin * 3, global_B_box[1][2] + margin * 4), a_min=0,
+                              a_max=256)  # leave more margins at the bottom
+    else:
+        # print(file_name)
+
+        DL_failed_detect_flag = True
+
+    if DL_failed_detect_flag:
+        img_name = os.path.join(out_folder_mismatched_image, file_name)
+        if not finding_lungs_non_DL_approach_and_save(raw_img, file_name):
+            # save file name only if both methods are not detected
+            image_misdetect_list.append(file_name)
+            print(file_name)
+    else:
+        img_name = os.path.join(out_folder_matched_img, file_name)
+        cropped = cv2.resize(raw_img, im_shape)[top_edge:bottom_edge, left_edge:right_edge]
+        # print(cropped)
+        resized_cropped = cv2.resize(cropped, im_shape)
+        cv2.imwrite(img_name, resized_cropped)
+
+    # if mis_detected_flag:
+    #     mis_detected_flag = False
+    #     fig, ax = plt.subplots(2, 2, figsize=[12, 12])
+    #     ax[0, 0].set_title("Original " + file_name)
+    #     ax[0, 0].imshow(raw_img, cmap='gray')
+    #     ax[0, 0].axis('off')
+    #     ax[0, 1].set_title("Threshold " + file_name)
+    #     ax[0, 1].imshow(thresh_img, cmap='gray')
+    #     #         ax[0, 1].imshow(prediction, cmap='gray')
+    #     ax[0, 1].axis('off')
+    #     ax[1, 0].set_title("Color Labels " + file_name)
+    #     ax[1, 0].imshow(labels)
+    #     ax[1, 0].axis('off')
+    #     ax[1, 1].set_title("Apply Mask on Original " + file_name)
+    #
+    #     ax[1, 1].imshow(resized_cropped, cmap='gray')
+    #     ax[1, 1].axis('off')
+
+    if index > 112120:  # for debug purpose
+        break
+
+    if index % 100 == 0:
+        df = pd.DataFrame({'col': image_misdetect_list})
+        df.to_csv(mis_detected_csv_path, header=False, index=False)
+
+df = pd.DataFrame({'col': image_misdetect_list})
+df.to_csv(mis_detected_csv_path, header=False, index=False)
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/non_PA_AP_view_samples.png b/Predict-Lung-Disease-master/src/finding_lungs/non_PA_AP_view_samples.png
new file mode 100644
index 0000000..0d140a5
Binary files /dev/null and b/Predict-Lung-Disease-master/src/finding_lungs/non_PA_AP_view_samples.png differ
diff --git a/Predict-Lung-Disease-master/src/finding_lungs/rotated_images_samples.png b/Predict-Lung-Disease-master/src/finding_lungs/rotated_images_samples.png
new file mode 100644
index 0000000..618f4f8
Binary files /dev/null and b/Predict-Lung-Disease-master/src/finding_lungs/rotated_images_samples.png differ
diff --git a/Predict-Lung-Disease-master/src/score_image_and_cam.py b/Predict-Lung-Disease-master/src/score_image_and_cam.py
new file mode 100644
index 0000000..ca1b40a
--- /dev/null
+++ b/Predict-Lung-Disease-master/src/score_image_and_cam.py
@@ -0,0 +1,215 @@
+# This script generates the scoring and schema files
+# Creates the schema, and holds the init and run functions needed to 
+# operationalize the chestXray model
+
+
+import os, sys, pickle, base64
+import keras.models
+import keras.layers
+import keras_contrib.applications.densenet
+import pandas as pd
+import numpy as np
+import azure_chestxray_utils, azure_chestxray_cam
+
+####################################
+# Parameters
+####################################
+global chest_XRay_model
+global as_string_b64encoded_pickled_data_column_name
+as_string_b64encoded_pickled_data_column_name   = 'encoded_image'
+global densenet_weights_file_name
+# densenet_weights_file_name = 'weights_only_chestxray_model_14_weights_712split_epoch_029_val_loss_147.7599.hdf5'
+densenet_weights_file_name = 'weights_only_chestxray_model_14_weights_712split_epoch_029_val_loss_147.7599 - Copy.hdf5'
+
+# Import data collection library. Only supported for docker mode.
+# Functionality will be ignored when package isn't found
+try:
+    from azureml.datacollector import ModelDataCollector
+except ImportError:
+    print("Data collection is currently only supported in docker mode. May be disabled for local mode.")
+    # Mocking out model data collector functionality
+    class ModelDataCollector(object):
+        def nop(*args, **kw): pass
+        def __getattr__(self, _): return self.nop
+        def __init__(self, *args, **kw): return None
+    pass
+
+####################################
+# Utils
+####################################
+def as_string_b64encoded_pickled(input_object):
+     #b64encode returns bytes class, make it string by calling .decode('utf-8')
+     return (base64.b64encode(pickle.dumps(input_object))).decode('utf-8')
+
+def unpickled_b64decoded_as_bytes(input_object):
+    if input_object.startswith('b\''):
+        input_object = input_object[2:-1]
+    # make string bytes
+    input_object   =  input_object.encode('utf-8')
+    #decode and the unpickle the bytes to recover original object
+    return (pickle.loads(base64.b64decode(input_object)))
+
+def get_image_score_and_serialized_cam(crt_cv2_image, crt_chest_XRay_model):
+    prj_consts = azure_chestxray_utils.chestxray_consts()
+    crt_cv2_image = azure_chestxray_utils.normalize_nd_array(crt_cv2_image)
+    crt_cv2_image = 255*crt_cv2_image
+    crt_cv2_image=crt_cv2_image.astype('uint8')
+    predictions, cam_image, predicted_disease_index = \
+    azure_chestxray_cam.get_score_and_cam_picture(crt_cv2_image, crt_chest_XRay_model)
+    blended_image = azure_chestxray_cam.process_cam_image(cam_image, crt_cv2_image)
+    serialized_image = azure_chestxray_cam.plot_cam_results(blended_image, cam_image, crt_cv2_image, \
+                 prj_consts.DISEASE_list[predicted_disease_index])
+    return predictions, serialized_image
+
+####################################
+# API functions
+####################################
+
+# Prepare the web service definition by authoring
+# init() and run() functions. Test the functions
+# before deploying the web service.
+def init():
+    try: 
+        print("init() method: Python version: " + str(sys.version))
+        print("crt Dir: " + os.getcwd())
+        
+        import pip
+        # pip.get_installed_distributions()
+        myDistr = pip.get_installed_distributions()
+        type(myDistr)
+        for crtDist in myDistr:
+            print(crtDist)
+
+        # load the model file
+        global chest_XRay_model
+        chest_XRay_model = azure_chestxray_utils.build_DenseNetImageNet201_model() 
+        chest_XRay_model.load_weights(densenet_weights_file_name)
+        print('Densenet model loaded')
+        
+    except Exception as e:
+        print("Exception in init:")
+        print(str(e))
+
+def run(input_df):
+    try:
+        import json
+        
+        debugCounter = 0
+        print("run() method: Python version: " + str(sys.version) ); print('Step '+str(debugCounter));debugCounter+=1
+
+        print ('\ninput_df shape {}'.format(input_df.shape))
+        print(list(input_df))
+        print(input_df)
+
+        input_df = input_df[as_string_b64encoded_pickled_data_column_name][0]; print('Step '+str(debugCounter));debugCounter+=1
+        input_cv2_image = unpickled_b64decoded_as_bytes(input_df); print('Step '+str(debugCounter));debugCounter+=1
+
+        #finally scoring
+        predictions, serialized_cam_image = get_image_score_and_serialized_cam(input_cv2_image, chest_XRay_model)
+        #predictions = chest_XRay_model.predict(input_cv2_image[None,:,:,:])
+
+        # prediction_dc.collect(ADScores)
+        outDict = {"chestXrayScore": str(predictions), "chestXrayCAM":as_string_b64encoded_pickled(serialized_cam_image)}
+        return json.dumps(outDict)
+    except Exception as e:
+        return(str(e))
+
+
+####################################
+# main function can be used for test and demo
+####################################
+def main():
+    from azureml.api.schema.dataTypes import DataTypes
+    from azureml.api.schema.sampleDefinition import SampleDefinition
+    from azureml.api.realtime.services import generate_schema
+
+    print('Entered main function:')
+    print(os.getcwd())
+    
+    amlWBSharedDir = os.environ['AZUREML_NATIVE_SHARE_DIRECTORY'] 
+    print(amlWBSharedDir)
+
+    def get_files_in_dir(crt_dir):
+        return( [f for f in os.listdir(crt_dir) if os.path.isfile(os.path.join(crt_dir, f))])
+
+    fully_trained_weights_dir=os.path.join(
+        amlWBSharedDir,
+        os.path.join(*(['chestxray', 'output',  'trained_models_weights'])))
+    crt_models = get_files_in_dir(fully_trained_weights_dir)
+    print(fully_trained_weights_dir)
+    print(crt_models)
+
+    test_images_dir=os.path.join(
+        amlWBSharedDir, 
+        os.path.join(*(['chestxray', 'data', 'ChestX-ray8', 'test_images'])))
+    test_images = get_files_in_dir(test_images_dir)
+    print(test_images_dir)
+    print(len(test_images))
+
+    # score in local mode (i.e. here in main function)
+    model = azure_chestxray_utils.build_DenseNetImageNet201_model()
+    model.load_weights(os.path.join(
+        fully_trained_weights_dir, densenet_weights_file_name))
+
+    print('Model weoghts loaded!')
+
+    import cv2
+    cv2_image = cv2.imread(os.path.join(test_images_dir,test_images[0]))
+    x, serialized_cam_image = get_image_score_and_serialized_cam(cv2_image, model)
+    file_bytes = np.asarray(bytearray(serialized_cam_image.read()), dtype=np.uint8)
+    recovered_image = cv2.imdecode(file_bytes, cv2.IMREAD_COLOR)
+
+    # x = model.predict(cv2_image[None,:,:,:])
+    print(test_images[0])
+    print(x)
+    print(recovered_image.shape)
+
+    #  score in local mode (i.e. here in main function) using encoded data
+    encoded_image = as_string_b64encoded_pickled(cv2_image)
+    df_for_api = pd.DataFrame(data=[[encoded_image]], columns=[as_string_b64encoded_pickled_data_column_name])
+    del encoded_image 
+    del cv2_image
+    del serialized_cam_image
+    
+    input_df = df_for_api[as_string_b64encoded_pickled_data_column_name][0]
+    input_cv2_image = unpickled_b64decoded_as_bytes(input_df); 
+    x, serialized_cam_image = get_image_score_and_serialized_cam(input_cv2_image, model) 
+    file_bytes = np.asarray(bytearray(serialized_cam_image.read()), dtype=np.uint8)
+    recovered_image = cv2.imdecode(file_bytes, cv2.IMREAD_COLOR)
+
+    # x = model.predict(input_cv2_image[None,:,:,:])
+    print('After encoding and decoding:')
+    print(x)
+    print(recovered_image.shape)
+
+    del model
+
+    # now create the post deployment env, i.e. score using init() and run()
+    crt_dir = os.getcwd()
+    working_dir = os.path.join(crt_dir, 'tmp_cam_deploy')
+    if not os.path.exists(working_dir):
+        os.makedirs(working_dir)
+
+    import shutil
+    shutil.copyfile(
+        os.path.join( fully_trained_weights_dir,densenet_weights_file_name), 
+        os.path.join( working_dir,densenet_weights_file_name)) 
+
+    os.chdir(working_dir)
+
+    # Turn on data collection debug mode to view output in stdout
+    os.environ["AML_MODEL_DC_DEBUG"] = 'true'
+
+    # Test the output of the functions
+    init()
+    print("Result: " + run(df_for_api))
+
+     # #Generate the schema
+    data_for_schema = {"input_df": SampleDefinition(DataTypes.PANDAS, df_for_api)}
+    schema_file = os.path.join(fully_trained_weights_dir, 'chest_XRay_cam_service_schema.json')
+    generate_schema(run_func=run, inputs=data_for_schema, filepath=schema_file)
+    print("Schema saved in " +schema_file)   
+    
+
+if __name__ == "__main__":
+    main()
diff --git a/_posts/README.md b/_posts/README.md
new file mode 100644
index 0000000..39bddb5
--- /dev/null
+++ b/_posts/README.md
@@ -0,0 +1,74 @@
+# Predict-Lung-Disease-through-Chest-X-Ray
+We obtain this repository by refactoring the [code](https://github.com/Azure/AzureChestXRay) for the blog post [Using Microsoft AI to Build a Lung-Disease Prediction Model using Chest X-Ray Images](https://blogs.technet.microsoft.com/machinelearning/2018/03/07/using-microsoft-ai-to-build-a-lung-disease-prediction-model-using-chest-x-ray-images/). This instruction aims to help newcomers build the system in a very short time.   
+# Installation
+1. Clone this repository
+   ```Shell
+   git clone https://github.com/svishwa/crowdcount-mcnn.git
+   ```
+   We'll call the directory that you cloned PredictLungDisease `ROOT`  
+  
+2. All essential dependencies should be installed：pickle, random, re, tqdm, cv2, numpy, pandas, sklearn, keras, tensorflow, keras_contrib, collections.counter.
+   
+# Data set up
+1. Download the NIH Chest X-ray Dataset from here:  
+   https://nihcc.app.box.com/v/ChestXray-NIHCC.  
+   You need to get all the image files (all the files under `images` folder in NIH Dataset), `Data_Entry_2017.csv` file, as well as the      Bounding Box data `BBox_List_2017.csv`.  
+
+2. Create Directory 
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC_other
+   ```  
+3. Save all images under `ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC`  
+
+4. Save `Data_Entry_2017.csv` and `BBox_List_2017.csv` under `ROOT/azure-share/chestxray/data/ChestX-ray8/ChestXray-NIHCC_other`  
+
+5. Process the Data
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/output/data_partitions
+   ```  
+   Run `000_preprocess.py` to create `*.pickle` files under this directory 
+# Test  
+1. We have provided the pretrained-model `azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5` under `ROOT/azure-share/chestxray/output/fully_trained_models`. You can also download it separately from [here](https://chestxray.blob.core.windows.net/chestxraytutorial/tutorial_xray/chexray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5).  
+
+2. Run `020_evaluate.py` and it will create `weights_only_azure_chest_xray_14_weights_712split_epoch_054_val_loss_191.2588.hdf5` saving weights of the pretrained-model under the same directory.
+
+3. Below is the result showing the AUC score of all the 14 diseases:  
+
+   | Disease            | Our AUC Score    | Stanford AUC Score | Delta     
+   |--------------------|------------------|--------------------|-----------:
+   | Atelectasis        | 0.822334         | 0.8094             | -0.012934 
+   | Cardiomegaly       | 0.933610         | 0.9248             | -0.008810 
+   | Effusion           | 0.882471         | 0.8638             | -0.018671 
+   | Infiltration       | 0.744504         | 0.7345             | -0.010004 
+   | Mass               | 0.858467         | 0.8676             |  0.009133 
+   | Nodule             | 0.784230         | 0.7802             | -0.004030 
+   | Pneumonia          | 0.800054         | 0.7680             | -0.032054 
+   | Pneumothorax       | 0.829764         | 0.8887             |  0.058936 
+   | Consolidation      | 0.811969         | 0.7901             | -0.021869 
+   | Edema              | 0.894102         | 0.8878             | -0.006302 
+   | Emphysema          | 0.847477         | 0.9371             |  0.089623
+   | Fibrosis           | 0.882602         | 0.8047             | -0.077902 
+   | Pleural Thickening | 1.000000         | 0.8062             | -0.193800 
+   | Hernia             | 0.916610         | 0.9164             | -0.000210   
+   
+# Visualization    
+1. Create Folder Test
+   ```Shell
+   mkdir ROOT/azure-share/chestxray/data/ChestX-ray8/test_images
+   ```  
+   Copy any number of images under `ChestXray-NIHCC` to `test_images` and resize them to 224x224 pixels.  
+
+2. Run `004_cam_simple.py` and it will output a Class Activation Map(CAM). The CAM lets us see which regions in the image were relevant to this class.  
+
+   ![这里随便写文字](https://github.com/fatLime/Predict-Lung-Disease/blob/master/image.png)
+
+# Referenced Paper  
+- Baseline result: https://arxiv.org/abs/1705.02315
+- Image Localization: http://arxiv.org/abs/1512.04150   
+- The original chexnet paper mentioned in [StanfordML website](https://stanfordmlgroup.github.io/projects/chexnet/) as well as their [paper](https://arxiv.org/abs/1711.05225).
+- http://cs231n.stanford.edu/reports/2017/pdfs/527.pdf for pre-processing the data
+- https://arxiv.org/abs/1711.08760 for some other thoughts on the model architecture and the relationship between different diseases
+ 
+# Notes
+  Please contact yanhaotian@bupt.edu.cn if you have any problem.