Multiparametric computer-aided differential diagnosis of Alzheimer’s disease and frontotemporal dementia using structural and advanced MRI

Abstract

ObjectivesTo investigate the added diagnostic value of arterial spin labelling (ASL) and diffusion tensor imaging (DTI) to structural MRI for computer-aided classification of Alzheimer's disease (AD), frontotemporal dementia (FTD), and controls.MethodsThis retrospective study used MRI data from 24 early-onset AD and 33 early-onset FTD patients and 34 controls (CN). Classification was based on voxel-wise feature maps derived from structural MRI, ASL, and DTI. Support vector machines (SVMs) were trained to classify AD versus CN (AD-CN), FTD-CN, AD-FTD, and AD-FTD-CN (multi-class). Classification performance was assessed by the area under the receiver-operating-characteristic curve (AUC) and accuracy. Using SVM significance maps, we analysed contributions of brain regions.ResultsCombining ASL and DTI with structural MRI resulted in higher classification performance for differential diagnosis of AD and FTD (AUC = 84%; p = 0.05) than using structural MRI by itself (AUC = 72%). The performance of ASL and DTI themselves did not improve over structural MRI. The classifications were driven by different brain regions for ASL and DTI than for structural MRI, suggesting complementary information.ConclusionsASL and DTI are promising additions to structural MRI for classification of early-onset AD, early-onset FTD, and controls, and may improve the computer-aided differential diagnosis on a single-subject level.Key points• Multiparametric MRI is promising for computer-aided diagnosis of early-onset AD and FTD. • Diagnosis is driven by different brain regions when using different MRI methods. • Combining structural MRI, ASL, and DTI may improve differential diagnosis of dementia. Electronic supplementary materialThe online version of this article (doi:10.1007/s00330-016-4691-x) contains supplementary material, which is available to authorized users.

Single Subject Classification of Alzheimer’s Disease and Behavioral Variant Frontotemporal Dementia Using Anatomical, Diffusion Tensor, and Resting-State Functional Magnetic Resonance Imaging

“…We speculate that other algorithms that utilize more sophisticated feature combination approaches, like sparse group lasso models [61], or hierarchical or longitudinal algorithms that aim to differentiate patients from a general population in order to subsequently differentiate between dementia-types may further exploit and weigh the additional information from multiple measures [64]. Incorporating other or additional imaging-derived biomarkers as cerebral blood flow [65], amplitude of low frequency fluctuations [32], GM derived connectomics [19], or diffusion tractography derived graph-based analytics [61] may further contribute to MRI-based dementiatype classification estimates without increasing diagnostic complexity.…”

Detection of mild cognitive impairment in a community‐dwelling population using quantitative, multiparametric MRI‐based classification

“…Previous MRI‐based classification methods have been heralded as promising tools for accurate classification of AD (Bouts et al, ; Bron et al, ; Schouten et al, ), MCI (Cui et al, ; Eskildsen et al, ), or to differentiate between MCI subjects likely to develop dementia due to AD or those that do not progress (Adaszewski et al, ; Arbabshirani, Plis, Sui, & Calhoun, ; Eskildsen et al, ; Misra et al, ; Rathore et al, ). These studies generally aimed to maximize classification performance by using sparse, carefully selected clinical samples.…”

“…While the other models used a separate clinical cohort for training. Nevertheless, it was previously observed that MCI detection models that used DTI‐derived measures (Dyrba et al, ) or combinations with measures of GM atrophy were best for the detection of MCI (Cui et al, ; Fan et al, ) or AD (Bron et al, ; Rathore et al, ; Schouten et al, ). We also found that only those models that either used DTI‐derived measures of impaired WM integrity or combined these with measures of GM atrophy were better than chance for MCI detection within the RS cohort.…”

“…Classification performance rates of the MCI‐model did not improve over those of the AD‐model or mild‐AD‐model. The incorporation of additional prior, biological knowledge (Rathore et al, ), or other information derived from imaging‐ and nonimaging biomarkers such as cerebrovascular status (De Bruijn et al, ), the load (De Bruijn et al, ; Dong et al, ; Fan et al, ), or location of white matter hyperintensities (McAleese et al, ), cerebral blood flow (Bron et al, ), resting‐state functional MRI (de Vos et al, ; Schouten et al, ), PET‐derived biomarkers (Dukart et al, ; Li et al, ), or additional cognitive assessment scores including measures of cognitive reserve (Allegri et al, ; Moradi et al, ; Vieira, Pinaya, & Mechelli, ; Wang et al, ) may further augment classification accuracy without increasing diagnostic complexity. While cognitive assessment scores would most likely provide a valuable contribution to the detection of MCI (Moradi et al, ; Wang et al, ), we did not consider these for this study.…”

“…Previous MRI-based classification methods have been heralded as promising tools for accurate classification of AD Bron et al, 2016;Schouten et al, 2016), MCI (Cui et al, 2012; Mean, minimum, and maximum area under the ROC curve (AUC) after 100 classification repetitions. Classifications with the AD-, mild-AD-, and moderate-AD-models resulted from 100 times repeated training on the AD cohort and applying it to the RS cohort.…”