To accomplish prostate lesion detection, pixel-level labels are manually annotated prior to model training. Prostate lesions are commonly marked with point label only, as fine delineating lesion region (e.g., the pixel-level label) is tedious. However, the point label, typically perceived as a weak label, is insufficient to represent the lesion area for segmentation model training. The lesion area not marked with the point label is probably categorized as negative (health tissue) pixel samples. Thus, we strengthen the existing “weak” point labels by aggregating its neighbor pixels into a region, providing promising cues for lesion detection. Kiraly, Abi Nader, Tuysuzoglu, Grimm, Kiefer, El-Zehiry and Kamen [28] expanded the single marked pixel to a small-diameter circle using Gaussian kernels. However, such a processing method focuses on lesion localization rather than contour approximation. Therefore, we apply a more sophisticated weak label processing method, i.e. distance regularized level set evolution [44], to automatically generate the coarse mask label (in Figure 4). This level set method is an edge-based active contour approach. The label can be produced in three steps: 1) Initialize a level set function to represent the lesion contour originated from a manually marked point; 2) Expand the lesion contour outward and update the level set function; 3) Terminate the expansion and finalize the function once exceeding the pre-defined iteration steps. As a result, the coarse mask label can be generated without labor-intensive annotation on the lesion region.