Journal of imaging informatics in medicine · 2026
Performance Analysis of Laplacian Re-Decomposition-Based PET-MRI Fusion for Quantitative Imaging Biomarkers in Alzheimer's Disease Classification.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Moslehi Masoud, Sharma Rahul, Khorasani Amir, Alzheimer’s Disease NeuroimagingInitiative
Аннотация
BACKGROUND: Alzheimer's disease (AD) and mild cognitive impairment (MCI) represent critical stages along the continuum of cognitive decline. This study investigates the Laplacian Re-decomposition (LRD) method for fusing positron emission tomography (PET) and magnetic resonance imaging (MRI) to develop a quantitative imaging biomarker for the classification of AD.
MATERIAL AND METHOD: Data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database were used in this study. T1, T2, T2-FLAIR, and T2*-weighted MRI sequences, together with [18F]FDG-PET images, were preprocessed and fused using the LRD algorithm. Hippocampal and frontal lobe WM segmentation, performed via U-Net, enabled relative signal contrast (RSC) calculation as a quantitative biomarker for each modality and fused image. Fusion and segmentation performance were assessed using standard metrics. RSC statistically compared between diagnostic groups after adjustment for demographic confounds, and evaluated for classification performance using ROC/AUC analysis.
RESULTS: The mean Dice coefficient achieved by the trained U-Net for hippocampus and frontal lobe WM segmentation were 0.93 and 0.95, respectively. ROC analysis revealed that the RSC extracted from T1-weighted images fused with PET outperformed other modalities for AD vs. Cognitively normal (CN) classification (AUC = 0.88, accuracy = 0.90). Also, T2-weighted images fused with PET, using RSC as a quantitative biomarker, achieved the best classification performance for AD vs. MCI (AUC = 0.81, accuracy = 0.81) and MCI vs. CN (AUC = 0.84, accuracy = 0.88) classification.
CONCLUSION: Our findings indicate that RSC extracted from LRD-fused images achieves competitive classification performance, supporting its potential as a quantitative biomarker for AD.
MATERIAL AND METHOD: Data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database were used in this study. T1, T2, T2-FLAIR, and T2*-weighted MRI sequences, together with [18F]FDG-PET images, were preprocessed and fused using the LRD algorithm. Hippocampal and frontal lobe WM segmentation, performed via U-Net, enabled relative signal contrast (RSC) calculation as a quantitative biomarker for each modality and fused image. Fusion and segmentation performance were assessed using standard metrics. RSC statistically compared between diagnostic groups after adjustment for demographic confounds, and evaluated for classification performance using ROC/AUC analysis.
RESULTS: The mean Dice coefficient achieved by the trained U-Net for hippocampus and frontal lobe WM segmentation were 0.93 and 0.95, respectively. ROC analysis revealed that the RSC extracted from T1-weighted images fused with PET outperformed other modalities for AD vs. Cognitively normal (CN) classification (AUC = 0.88, accuracy = 0.90). Also, T2-weighted images fused with PET, using RSC as a quantitative biomarker, achieved the best classification performance for AD vs. MCI (AUC = 0.81, accuracy = 0.81) and MCI vs. CN (AUC = 0.84, accuracy = 0.88) classification.
CONCLUSION: Our findings indicate that RSC extracted from LRD-fused images achieves competitive classification performance, supporting its potential as a quantitative biomarker for AD.