Computer Aided Diagnosis System of Meniscal Tears with T1 and T2 Weighted MR Images Based on Fuzzy Inference

摘要

This paper proposes a computer aided diagnosis system of meniscal tears from 3D human knee MR image with Tl-weighted and T2-weighted MR images. The first step of our method is a 3D image registration between both images on the computer display by manual. The second step determines the candidate region of the menisci from T2-weighted MR image aided by fuzzy if-then rules with respect to the location and the intensity. The final step determines the voxels in the menisci from the candidate region by using T1-weighted and T2-weighted MR images. We applied this method to several subjects. All voxels in the menisci of each subject were successfully identified and their 3D surfaces were displayed. Thus, our developed system would improve to diagnose the meniscal tears.