When it comes to applying computer vision in the medical field, most tasks involve either 1) image classification for diagnosis or 2) segmentation to identify and separate lesions. However, in pathology cancer detection, this is not always possible. Obtaining labels is time-consuming and labor-intensive. Furthermore, pathology slides can be up to 200k x 100k pixels resolution, and they will not fit in memory for classification since for example, the ImageNet only uses 224 x 224 pixels for training. Downsampling normally is not an option because we are trying to detect a tiny area, such as a cancerous area varying from 300 x 300 pixels area (a few dots in Figure 1).
Sep-17-2020, 01:50:23 GMT