Automated image based registration of endoscopy to CT views for improved target delineation in image guided radiation therapy
Souzan Mirza · TSpace (University of Toronto) · 2018
One of the largest uncertainties in radiotherapy treatment planning for squamous cell carcinomas of the head and neck is contouring the gross tumour volume (GTV). The treatment planning CT is used to contour the GTV, but under-contouring of the mucosal extent of the disease is common. Endoscopy exams can be used to reduce GTV errors by registering endoscopy images to the CT to project contours of the mucosal extent of the disease viewed under endoscopy. To register endoscopy and CT images, a rigid registration pipeline was developed using electromagnetic tracking and image based registration with a target of achieving a contouring error of less than or equal to 2mm between the two image sets. On phantom data, a contouring error of less than or equal to 2mm was achieved in 14 of 18 images. On four images of glottal tumours from two clinical cases, the contouring error ranged between 2.4-6.1mm.