Analysis of MRI as a screening tool for the diagnosis of schizophrenia

• Published in: Journal of physics. Conference series Vol. 2318; no. 1; pp. 12036 - 12047
• Main Authors: Thilakavathi, B, Sudha, S, Vidya, K, Subathra, Y
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.08.2022
• Subjects: Back propagation networks; Classification; Datasets; Euclidean distance; Euclidean geometry; Hausdorff dimension; Magnetic resonance imaging; MRI; Neural networks; Physics; Schizophrenia; Schizophrenia (SCZ)
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/2318/1/012036

Abstract Schizophrenia (SCZ) is a clinical disorder that affects 0.01% of the world population. It affects people in late teen hood or early maturity resulting in lifelong social and mental disturbance. At present, there is no cure, but this can be diagnosed and treated. Classification of SCZ imposes great challenges even for the most experienced neurologists. A non-intrusive technique like MRI is taken for diagnosing various diseases which are used as a base for our tool. Many researchers used large datasets of SCZ and normal for analyzing SCZ using various parameters like Grey matter, white matter, voxel-based morphometry, etc., This work proposes a simpler but effective approach to classify the same. This paper determined statistical and complexity features from 32 SCZ and 18 normal MRI images. Totally 9 features are determined out of these, novel features Hausdorff dimension and Euclidean distance played an important role in classification. Hausdorff dimension is selected as the most significant feature by student’s t-test with p<0.001 . The back propagation neural network receives substantial information from the t-test as input. Our promising approach, with a minimal dataset, classified the subjects with 100% sensitivity, 88.9% specificity, and 94.4% accuracy.