Shaji S, Palanisamy R, Swaminathan R. Explainable Optimized LightGBM Based Differentiation of Mild Cognitive Impairment Using MR Radiomic Features.
Stud Health Technol Inform 2022;
295:483-486. [PMID:
35773916 DOI:
10.3233/shti220770]
[Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 06/15/2023]
Abstract
In this study, an explainable Bayesian Optimized (BO) LightGBM model is employed to differentiate the Corpus Callosal (CC) image features of Healthy Controls (HC) and Mild Cognitive Impairment (MCI). For this, Magnetic Resonance (MR) brain images obtained from a public database are pre-processed and CC is segmented using spatial fuzzy clustering-based level set. Radiomic features are extracted from the segmented CC, which are further fed to BO-LightGBM classifier. SHapley Additive exPlanations (SHAP) technique is used to evaluate the interpretability of the model. The results indicate that radiomics based BO-LightGBM is able to differentiate MCI from HC. An area under curve of 0.83 is achieved by the model. SHAP values suggest that out of 56 radiomic features, texture descriptors possess the highest discriminative power in MCI diagnosis. The performance of adopted approach indicates that radiomics based BO-LightGBM aid in the automated diagnosis of early Alzheimer's Disease stages.
Collapse