Fig. 2

Clips segmented by the proposed model with different feature inputs. (a) The model output without both binary image and feature map, (b) The model output with binary image and without feature map input, (c) The model output without binary image and with feature map input, (d) The model output with both binary image and feature map, and (e) True clip contour delineated manually by physician. The red areas indicate the regions detected by the proposed models or delineated by physician