A New Adaptation Phase for Thresholds in a CBR System Associated to a Region Growing Algorithm to Segment Tumoral Kidneys

Image segmentation is an abundant topic for computer vision and image processing. Most of the time, segmentation is not fully automated, and a user is required to guide the process in order to obtain correct results. Yet, even with programs, it is a time-consuming process. In a medical context, segmentation can provide a lot of information to surgeons, but since this task is manual, it is rarely executed because of time. Artificial Intelligence (AI) is a powerful approach to create viable solutions for automated treatments. In this paper, we reused a case-based reasoning (CBR) system previously developed to segment renal parenchyma with a region growing algorithm and we completed its adaptation phase allowing a better adjustment of parameters before segmentation. Compared to the previous system, we added an adaptation for the thresholds values in addition to the adaptation of the seeds coordinates. We compared several versions of our new adaptation in order to determine the best and we confronted it with a deep learning approach realized in similar conditions.