Image enhancement techniques in an image monitoring system for total knee arthroplasty

摘要

An image-guided surgical navigation system for Total Knee Arthroplasty (TKA) has been developed. It provides internal images of the knee joint during TKA surgeries, and help surgeons to decide the proper position of implant components. The images are captured under extremely low illumination, and image quality enhancement is needed for better judgment. For this purpose, we propose a two-step image processing pipeline which consists of the denoising and the contrast enhancement. The BM3D denoising method improves the peak signal-to-noise ratio by +12dB. A modified Retinex algorithm is developed for contrast enhancement, and improves the contrast-per-pixel of original internal knee images by 59.12% on average and improves the discrete entropy by 23.52%. Experiments on images taken under the emulated environment have been taken to demonstrate the effectiveness.