Fractal dimension: analyzing its potential as a neuroimaging biomarker for brain tumor diagnosis using machine learning

• DOI: 10.3389/fphys.2023.1201617
• Publication Date: 2023-07-18T01:15:33Z
• Authors: Dheerendranath Ba...; Sreejith Vidyadharan; B. V. V. S. N. Pr...; P. Yogeeswari; C. Kesavadas; Venkateswaran Raj...
• Keywords: Brain tumor

Purpose: The main purpose of this study was to comprehensively investigate the potential fractal dimension (FD) measures in discriminating brain gliomas into low-grade glioma (LGG) and high-grade (HGG) by examining tumor constituents non-tumorous gray matter (GM) white (WM) regions. Methods: Retrospective magnetic resonance imaging (MRI) data 42 patients (LGG, n = 27 HGG, 15) were used study. Using MRI, we calculated different FD based on general structure, boundary, skeleton aspects tumorous GM WM Texture features, namely, angular second moment, contrast, inverse difference correlation, entropy, also measured efficacy features assessed comparing them with texture features. Statistical inference machine learning approaches aforementioned distinguish LGG HGG patients. Results: from regions able Among 15 measures, structure values enhanced yielded high accuracy (93%), sensitivity (97%), specificity (98%), area under receiver operating characteristic curve (AUC) score (98%). Non-tumorous good (83.3%), (100%), (60%), AUC (80%) classifying grades. These found be significantly (p < 0.05) between On other hand, among 25 region revealed significant differences HGG. In learning, accuracy, sensitivity, specificity, score. Conclusion: A comparison that analysis components not only distinguished statistical significance classification but provided better insights grade classification. Therefore, can serve as neuroimaging biomarkers for glioma.