Explainable AI-based Deep-SHAP for mapping the multivariate relationships between regional neuroimaging biomarkers and cognition

• Published in: European journal of radiology Vol. 174; p. 111403
• Main Authors: Bhattarai, Puskar, Thakuri, Deepa Singh, Nie, Yuzheng, Chand, Ganesh B.
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.05.2024
• Subjects: Alzheimer Disease - diagnostic imaging; Alzheimer Disease - pathology; Amyloid-beta; Artificial Intelligence; Biomarkers; Brain - diagnostic imaging; Brain - pathology; Cognition; Cognitive Dysfunction - diagnostic imaging; Cognitive Dysfunction - pathology; Deep learning; Feature importance; Humans; Machine learning algorithms; Magnetic Resonance Imaging - methods; Mild cognitive impairment; Neuroimaging; Positron-Emission Tomography - methods; Shapley additive explanations; Tomography, X-Ray Computed; DVR; Machine learning algorithms; SHAP; Shapley additive explanations; AD; Mild cognitive impairment; MRI; PiB; AI; ROI; Deep learning; MAE; Feature importance; Aβ; MCI; OASIS; FDR; MMSE; PET; Amyloid-beta
• ISSN: 0720-048X; 1872-7727; 1872-7727
• DOI: 10.1016/j.ejrad.2024.111403

•Relationships among amyloid PET, MRI, and cognition are unclear in Alzheimer.•Deep-SHAP is proposed to investigate regional imaging vs. cognition relationships.•Deep-SHAP identified the modality-specific brain regions relevant to cognition. Mild cognitive impairment (MCI)/Alzheimer's disease (AD) is associated with cognitive decline beyond normal aging and linked to the alterations of brain volume quantified by magnetic resonance imaging (MRI) and amyloid-beta (Aβ) quantified by positron emission tomography (PET). Yet, the complex relationships between these regional imaging measures and cognition in MCI/AD remain unclear. Explainable artificial intelligence (AI) may uncover such relationships. We integrate the AI-based deep learning neural network and Shapley additive explanations (SHAP) approaches and introduce the Deep-SHAP method to investigate the multivariate relationships between regional imaging measures and cognition. After validating this approach on simulated data, we apply it to real experimental data from MCI/AD patients. Deep-SHAP significantly predicted cognition using simulated regional features and identified the ground-truth simulated regions as the most significant multivariate predictors. When applied to experimental MRI data, Deep-SHAP revealed that the insula, lateral occipital, medial frontal, temporal pole, and occipital fusiform gyrus are the primary contributors to global cognitive decline in MCI/AD. Furthermore, when applied to experimental amyloid Pittsburgh compound B (PiB)-PET data, Deep-SHAP identified the key brain regions for global cognitive decline in MCI/AD as the inferior temporal, parahippocampal, inferior frontal, supratemporal, and lateral frontal gray matter. Deep-SHAP method uncovered the multivariate relationships between regional brain features and cognition, offering insights into the most critical modality-specific brain regions involved in MCI/AD mechanisms.