Multi-phase object contour refinement

A method for providing lesion contour refinement for medical images, the method comprising:

• receiving a 3D input mask;
• receiving input slices from the multi-phase medical images, including a lesion, wherein the multi-phase medical images include at least one of a venous phase, an arterial phase, a pre-contrast phase, and a delayed phase for a same organ;
• cropping the input slices to remove a sub-portion of the input slices;
• performing lesion contour refinement, on one lesion at a time, for the cropped input slices and the 3D input mask to obtain a 3D predicted mask using a deep learning method;
• storing the 3D predicted mask that includes lesion contour refinement to obtain a 3D lesion contour;
• determining whether the 3D predicted mask meets a minimal overlap requirement;
• and
• only storing the 3D predicted mask when the 3D predicted mask meets the minimal overlap requirement.

A non-transitory computer medium including instructions that, when executed as a set of instructions by an electronic processor perform a set of operations comprising:

• receiving a 3D input mask;
• receiving input slices from the multi-phase medical images, including a lesion, wherein the multi-phase medical images include at least one of a venous phase, an arterial phase, a pre-contrast phase, and a delayed phase for a same organ;
• cropping the input slices to remove a sub-portion of the input slices;
• performing lesion contour refinement, on one lesion at a time, for the cropped input slices and the 3D input mask to obtain a 3D predicted mask using a deep learning method;
• storing the 3D predicted mask that includes lesion contour refinement to obtain a 3D lesion contour;
• determining whether the 3D predicted mask meets a minimal overlap requirement;
• and
• only storing the 3D predicted mask when the 3D predicted mask meets the minimal overlap requirement.