Automated registration of pre-operative head CT image and pathology images for osteoradionecrosis area estimation

Osteoradionecrosis of the jaw is the most refractory disease, which is caused by medical treatment such as radiotherapy and inhibitors of bone resorption. Recent researchers revealed that early resection of necrotic jawbone allows one to complete the treatment in a short period of time. The resection area in necrotic jawbone is determined by physician’s subjective diagnosis using pre-operative images. It takes a long time for the surgical planning and often differs from the intra-operative findings. This research aims to propose resection area of the jawbone accurately. Therefore, we automatically register pathological images to pre-operative head CT image. In the pre-operative head CT image and resected jawbone CT image registration, we proposed a two-step registration method using three manually acquired keypoints, and obtained normalized cross-correlation of 0.953 in the five subjects. The proposed two-step registration method achieved dice coefficient of 0.700 in the resected jawbone CT image and pathology image registration.