Head and Neck Cancer Tumor Segmentation Using Support Vector Machine in Dynamic Contrast-Enhanced MRI

Figure 3

Examples of tumor segmentation results from four patients versus the ground truth. Row-wise, from top to bottom, corresponding to 4 typical samples with AOM of 0.89, 0.79, 0.67, and 0.71, respectively. (a) Ground truth of the tumor drawn by an experienced radiologist (in red); (b) the blood vessels identified by the first classifier (in white); (c) the cavity identified by the second classifier (in white); (d) the normal tissue identified by the third classifier (in white); (e) the tumor region (in white) segmented by removing the voxels identified in (b), (c), and (d) from the region of interest (ROI).

(a)

(b)

(c)

(d)

(e)