Depth Sensor-Based Realtime Tumor Tracking for Accurate Radiation Therapy

Abstract

We present an image guided radiation therapy (IGRT) system for tracking tumors in realtime based on continuous structured light imaging. While an accurate positioning of the radiation isocenter to pre-imaged cancer cells is critical to minimize the risk of damaging healthy tissues, patients undergo involuntary motions such as breathing or unpredictable gestures during treatment. Moreover, multiple sessions are typically necessary and repositioning the patient accurately can be difficult. Our approach consists of determining the tumor position by densely tracking the deformation of a stream of 3D scans using a realtime variant of a state-of-the-art non-rigid registration algorithm and an FEM simulation on the interior body. We use interactive reprojection for visual guidance to adjust the posture of the patient and couch position, depending on the tumor location. Compared to existing techniques, our method uniquely estimates tumor deviations under body deformations. Our pipeline has been successfully commercialized as part of the C-RAD AB CatalystTM product line and is already deployed in a number of hospitals.