Detecting pancreatic masses on CT scans using YOLO architectures

This paper describes a methodology to aid in the localization and detection of masses in computed tomography scans. Cancerous tumors present in CT scans have a high similarity to the texture of the pancreas, making their diagnosis challenging. In addition, it is important to identify the tumor in the early stages, to increase the patient’s chances of survival. To assist in this task, a methodology is proposed using YOLO models to localize the pancreas and then locate the cancerous tumor in the previously located pancreas, with the aim of creating a subvolume that encompasses the tumor, facilitating diagnosis.