GI-Tract-Image-Segmentation

Description

In this project, we present our approaches to the Kaggle GI Tract Image Segmentation challenge. Since radiation oncologists try to deliver high doses of radiation using X-ray beams pointed to tumors while avoiding the stomach and intestines, the goal of the challenge is to effectively segment the stomach and intestines in MRI scans in order to improve the cancer treatment, circumventing the need for the time-consuming and labor intensive process in which radiation oncologists must manually outline the position of the stomach and intestines. We apply U-Net method to segment the organ areas. Our best U-Net model achieves a Jaccard Index of 0.96 on the validation set.

Dataset Example

Model Output