Figure 7.

The schematic diagram of segmented boundary reconstruction for the testing image. ① HVCMA, ② DCMA, ③ deep learning segmentation prediction from model 1 or mode 2, ④ extracting the tumor boundaries and mapping them to the original image coordinates, ⑤ transforming the rectangular coordinates to the polar coordinates and expands the data, ⑥ transforming the boundary to the rectangular coordinate system after boundary fitting and interception, ⑦ compensating and expanding the boundary data).