Multiple sclerosis (MS) is a type of auto-immune disease affecting the human brain. One of the challenges in clinical diagnosis of MS lies in its close similarity to another type of brain auto-immune disease known as neuromyelitis optica (MNO). Hence, various researchers are looking into the development of computer-aided differential diagnosis (CADD) of MS vs. MNO. Generally, the development of the CADD, involves image pre-processing, feature extraction, statistical analysis and training and testing of the classification model. The focus of this paper is on the 2 critical pre-processing steps prior to classification of MS vs. MNO, namely skull stripping and lesion segmentation. In particular, both skull stripping and brain lesion segmentation are tested with a popular semantic segmentation technique, DeepLabV3. For comparison, a recently published technique, SynthStrip from FreeSurfer 1, developed by Harvard University is also tested for brain extraction / skull removal. Similarly for segmentation of brain lesions, Lesion Segmentation Tool in SPM Toolbox is tested and compared. The results from this study provides insight on what are the appropriate pre-processing techniques for the development of an reliable automatic differential diagnosis tool for multiple sclerosis conditions. © 2023 IEEE.