Semi-automatic Segmentation of Sacrum in Computer Tomography Studies for Intraoperative Radiation Therapy

Abstract

Intra-Operative Radiation Therapy (IORT) is a technique that combines surgery and adjuvant radiation. It is applied directly to a post-resected tumor bed by means of a specific applicator docked to the linear accelerator. IORT’s primary purpose is to reduce local cancer recurrence. Separating healthy tissues from the tumor bed is essential in order to maximize treatment efficacy and to avoid damage to organs at risk. Current image segmentation during planning procedures is done manually, using computed tomography studies, by an expert radiation oncologist. This procedure can be time consuming. The authors present an automated procedure that would expedite the process of segmentation, specifically tested in rectal cancer cases where the sacrum is the area at risk. A segmentation algorithm was developed based on the creation of an averaged image (template) that was manually segmented. The algorithm geometrically transforms template segmentation by registering it to the input images. Preliminary results demonstrate the feasibility of the proposed approach.