Multiple sclerosis (MS) is a chronic demyelinating disease of the central nervous system whose diagnosis depends heavily on the reading of brain magnetic resonance imaging (MRI), a task that is time-consuming and subject to inter-reader variability. Published deep learning work on MS concentrates on lesion segmentation and classification algorithms, whereas the packaging of such a model into an auditable, deployable decision-support workflow with explicit evaluation safeguards receives less attention. This paper presents the design of Turing Body, a web-integrated framework that classifies brain MRI images as MS or normal using a convolutional neural network (CNN) and returns a report containing the predicted class, a confidence score and a lesion visualisation. The pipeline comprises slice extraction, resizing, intensity normalisation, noise reduction and data augmentation, followed by CNN classification implemented in Python with TensorFlow/Keras. The model is exposed through a Flask inference service, coordinated by a Spring Boot backend, presented through a React interface and persisted in MongoDB. We document the layered architecture, the functional and non-functional requirements, and the current implementation status, in which the backend and model modules are functional while frontend integration remains in progress. Because the project has not yet produced quantitative performance results, no accuracy figures are claimed. Instead, we define a patient-wise evaluation protocol, a metric set and a reporting standard intended to prevent slice-level data leakage and control-cohort confounding. The contribution is a documented reference architecture and a pre-specified, reproducible evaluation plan that subsequent experiments on public MS MRI benchmarks can follow.

This paper presented the design of Turing Body, a web-integrated CNN framework for MRI-based MS screening, covering its layered architecture, requirements, preprocessing pipeline and development process. An evidence audit showed that the backend and model modules are reported as functional, that the frontend remains in progress, and that no quantitative performance results exist yet. We therefore claim no accuracy and instead specify a patient-wise evaluation protocol that addresses slice-level leakage, control-cohort confounding and uncertainty. The principal contribution is a documented reference architecture and a pre-specified, reproducible evaluation plan. The next stage of the project is to execute that plan and report the findings alongside the design presented here.

[1]D. S. Reich, C. F. Lucchinetti, and P. A. Calabresi, “Multiple sclerosis,” N. Engl. J. Med., vol. 378, pp. 169–180, 2018. [2]A. J. Thompson et al., “Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria,” Lancet Neurol., vol. 17, no. 2, pp. 162–173, Feb. 2018, doi: 10.1016/S1474-4422(17)30470- 2. [3]A. Carass et al., “Longitudinal multiple sclerosis lesion segmentation: Resource and challenge,” NeuroImage, vol. 148, pp. 77–102, Mar. 2017, doi: 10.1016/j.neuroimage.2016.12.064. [4]A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” in Adv. Neural Inf. Process. Syst., vol. 25, 2012, pp. 1097–1105. [5]I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA, USA: MIT Press, 2016. [6]G. Litjens et al., “A survey on deep learning in medical image analysis,” Med. Image Anal., vol. 42, pp. 60–88, Dec. 2017, doi: 10.1016/j.media.2017.07.005. [7]A. Shoeibi et al., “Applications of deep learning techniques for automated multiple sclerosis detection using magnetic resonance imaging: A review,” Comput. Biol. Med., vol. 136, Art. no. 104697, Sep. 2021, doi: 10.1016/j.compbiomed.2021.104697. [8]O. Commowick et al., “Objective evaluation of multiple sclerosis lesion segmentation using a data management and processing infrastructure,” Sci. Rep., vol. 8, Art. no. 13650, Sep. 2018, doi: 10.1038/s41598-018-31911-7. [9]S. Valverde et al., “Improving automated multiple sclerosis lesion segmentation with a cascaded 3D convolutional neural network approach,” NeuroImage, vol. 155, pp. 159–168, 2017. [10]K. Kamnitsas et al., “Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation,” Med. Image Anal., vol. 36, pp. 61–78, 2017, doi: 10.1016/j.media.2016.10.004. [11]O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in Proc. Med. Image Comput. Comput.-Assist. Interv. (MICCAI), Cham, Switzerland: Springer, 2015, pp. 234–241, doi: 10.1007/978-3-319-24574-4_28. [12]K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Las Vegas, NV, USA, Jun. 2016, pp. 770–778, doi: 10.1109/CVPR.2016.90. [13]R. R. Selvaraju et al., “Grad-CAM: Visual explanations from deep networks via gradient-based localization,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), Venice, Italy, Oct. 2017, pp. 618–626, doi: 10.1109/ICCV.2017.74.