Diagnosis of optic neuritis using magnetic resonance images

Optic neuritis is an acute inflammation of myelin sheath that damages optic nerve while Magnetic Resonance Imaging (MRI) is one of the non-invasive alternatives to diagnose optic neuritis by measuring the mean cross-sectional area of the optic nerve. However, the extraction and analysis of optic nerve with MRI are challenging due to its discrete dimension and low spatial resolution of the MR images. This research leverages both image segmentation and interpolation to achieve better performance in MR image processing. The chosen image processing models are Level Set Method-Iterative Curvature Based Interpolation (LSM-ICBI) model and Reverse Diffusion-Level Set Method (RD-LSM) for T1 and T2 weighted images respectively. Both LSM-ICBI and RD-LSM models produce distinct optic nerve edges for the area measurement on the coronal view MR image slices. We compare the measurements of six datasets with the mean cross-sectional area of the normal optic nerves (27.51 +/- 0.83 mm(2) for T1 weighted image and 22.26 +/- 1.29 mm(2) for T2 weighted image). Our experimental results show that the accuracy of LSM-ICBI diagnosis for T1 weighted image is 83.33% while RD-LSM model achieves 66.67% in T2 weighted image.