The files in this folder are the test results on the CT image No.0 in the test data set of Medical Segmentation Decathlon (file imagesTr/liver_0.nii.gz in Task03_Liver.tar). The 3D images were displayed as videos along the Z axis.
- input.mp4: the raw CT image (selected part with liver only)
- labels.mp4: the groud truth segmentation label (mask) for this image
- predictions.mp4: the label prediction on this CT image by the trained network (along Z axis)
- pred_axis0.mp4: the predicted label along X axis
- pred_axis1.mp4: the predicted label along Y axis
- pred_final.mp4: the fused label of the three directions
Since the network was trained only on Z axis, the results for X and Y axes are not accurate. This fusion result is just an explanation of the idea.