Chaos theory for prognostic purposes in multiple sclerosis

Abstract

Multiple sclerosis is a chronic neuro-inflammatory disease. Its diagnosis and evaluation require a visual assessment in brain magnetic resonance images. This neurological disease is characterized by its unpredictable evolution. The most common form is relapsing–remitting multiple sclerosis, consisting of episodes of neurological dysfunction remitting with a variable degree of recovery. The underlying mechanisms are still unknown and have no mathematical models to explain this spatial and temporal dissemination. The main objective of this paper is to propose an approach based on chaos theory to define the clinical characteristics for lesion progression, principally its evolution. Previous authors have explored the nervous system by accurately modeling their attractors in the phase space and reproducing a non-linear result. This remains to be compared with chaotic attributes. In this work, multiple sclerosis lesions are treated through modeling and calculating of their degree of evolution. This is a retrospective study of cases collected from the National Institute of Neurology Mongi Ben Hmida. It included 74 patients with an age group ranging from 20 to 72 years. Four clinical trials are properly discussed: unstable cases with active lesions may be described by dynamic and non-linear systems presenting higher chaos value compared to ill patients with inactive lesions. On the other hand, the accuracy of chaos demonstration can be favorably affected by the reduced resolution of magnetic resonance images. Therefore, the interest is in preprocessing to improve the eventual results. High accuracy was achieved for the models discussed in this paper (91.97–85.1%). Accordingly, the suitability and practical usefulness of the ‘simple’ pretreatment to achieve multiple sclerosis classification are demonstrated.