A Novel Classification and Diagnosis of Multiple Sclerosis Method using Artificial Neural Networks and Improved Multi-Level Adaptive Conditional Random Fields

Due to the small size, low contrast, and variable position, shape, and texture of multiple sclerosis lesions, one of the challenges of medical image processing is the automatic diagnosis and segmentation of multiple sclerosis lesions in magnetic resonance images. Early diagnosis of these lesionns at the first stages of the disease can effectively diagnose and evaluate the treatment. Also automated segmentation is a powerful tool to assist the professionals in improving the accuracy of the disease diagnosis. In this work, we use the modified adaptive multi-level conditional random fields and the artificial neural network in order to segment and diagnose multiple sclerosis lesions. Instead of assuming the model coefficients as constant, they are considered as the variables in the multi-level statistical models. This work aims to evaluate the probability of lesions based on the severity, texture, and adjacent areas. The proposed method is applied to 130 MR images of the multiple sclerosis patients in two test stages, and results in a 98% precision. Also the proposed method reduces the error detection rate by correcting the lesion boundaries using the average intensity of neighborhoods, rotation invariant, and texture for very small voxels with a size of 3-5 voxels, and it shows very few false-positive lesions. The suggested model results in a high sensitivity of 91% with a false positive average of 0.5.