Active Contour Segmentation of the Anterior Cruciate Ligament in MR Images

Anterior Cruciate Ligament (ACL) injuries are common among athletes. The preferred source for diagnosis of ACL tears is Magnetic Resonance (MR) Imaging. Manually analyzing MR images is time-consuming and prone to error. In this study, an attempt has been made to segment ACL from the MR images using active contour-based segmentation. The images are sourced from Knee-MRI Dataset. The noises present in images are removed using Gaussian blur and median filters. Contrast enhancement is performed using histogram equalization, gamma correction and contrast stretching. Image sharpening is done by high-boost and Laplacian filters. The preprocessed images are subjected to active contour algorithm to segment ACL region and are validated with the ground-truth. The results demonstrate that median filter performed better for noise removal, gamma correction for contrast enhancement and Laplacian filter for image sharpening. Active contour segmentation along with morphological operations is able to segment the ACL regions in all the three classes, viz. non-injured, partially injured and completely ruptured with sensitivity of 88%, specificity of 99% and similarity of 82%. As early diagnosis of ACL tear is essential to prevent further damage, this approach could be clinically significant to improve the patient's health through timely intervention.