Classification of optic neuritis in neuromyelitis optica spectrum disorders (NMOSD) on MRI using CNN with transfer learning and manipulation of pre-processing on augmentation

Yang Feng,Li Sze Chow,Nadia Muhammad Gowdh,Norlisah Ramli,Li Kuo Tan,Suhailah Abdullah
DOI: https://doi.org/10.1088/2057-1976/ad6f17
2024-08-27
Abstract:Neuromyelitis optica spectrum disorder (NMOSD), also known as Devic disease, is an autoimmune central nervous system disorder in humans that commonly causes inflammatory demyelination in the optic nerves and spinal cord. Inflammation in the optic nerves is termed optic neuritis (ON). ON is a common clinical presentation; however, it is not necessarily present in all NMOSD patients. ON in NMOSD can be relapsing and result in severe vision loss. To the best of our knowledge, no study utilises deep learning to classify ON changes on MRI among patients with NMOSD. Therefore, this study aims to deploy eight state-of-the-art CNN models (Inception-v3, Inception-ResNet-v2, ResNet-101, Xception, ShuffleNet, DenseNet-201, MobileNet-v2, and EfficientNet-B0) with transfer learning to classify NMOSD patients with and without chronic ON using optic nerve magnetic resonance imaging. This study also investigated the effects of data augmentation before and after dataset splitting on cropped and whole images. Both quantitative and qualitative assessments (with Grad-Cam) were used to evaluate the performances of the CNN models. The Inception-v3 was identified as the best CNN model for classifying ON among NMOSD patients, with accuracy of 99.5%, sensitivity of 98.9%, specificity of 93.0%, precision of 100%, NPV of 99.0%, and F1-score of 99.4%. This study also demonstrated that the application of augmentation after dataset splitting could avoid information leaking into the testing datasets, hence producing more realistic and reliable results.
What problem does this paper attempt to address?