Fig. 3
From: Texture analysis on MR images helps predicting non-response to NAC in breast cancer
Automated segmentation of the tumor lesion. A rectangular area covering the breast is placed (a). Pixel-level calculation of parameters wash-in (b) and amplitude (c) is performed. Pixels are partitioned into k = 2 clusters (d). Morphological opening is applied to preserve the largest region of contiguous pixels with the same behavior in amplitude and wash-in only (e). Comparison with the manual delineation of the lesion shows an overall good agreement (f)