Differentiation of dementia with Lewy bodies from normal elderly and Alzheimer’s disease: A DLB Resemblance Atrophy Index

Abstract

AbstractBackgroundIt is sometimes challenging for clinicians to differentiate Dementia with Lewy Bodies (DLB) from normal elderly and Alzheimer’s disease (AD), particularly at the early stage. Since structural difference in multiple brain regions exists between DLB and normal elderly and AD, we aimed to derive a single severity index by combining these multiple MRI features from machine learning and explored the performance of this index (DLB resemblance atrophy index [DLB‐RAI]) in differentiating mild DLB from normal elderly and AD.Method61 DLB subjects, 128 AD subjects (mean clinical dementia rating [CDR]=0.852 vs 0.953, p=0.248) and 128 normal controls (NC) with age matched were recruited from the National Alzheimer’s Coordinating Center (NACC) database. All MRI images were postprocessed by an MRI‐based volumetry segmentation tool (AccuBrain®). We analyzed the whole brain structural difference between DLB and AD and NC subgroups. All brain features were utilized to derive a DLB‐RAI. We obtained the optimal cutoff of DLB‐RAI and used 10‐fold cross‐validation method to validate the performance of DLB‐RAI among the three subgroups.ResultThe optimal cutoff of DLB‐RAI in the differentiation of mild DLB from NC and AD was 0.5. The accuracy of DLB‐RAI (≥0.5) in distinguishing DLB from NC was 83.0% (AUC=0.87, Figure 1) with an optimal sensitivity of 0.83 and specificity of 0.86, while the accuracy in differentiating between DLB and AD was 75.10% (AUC=0.77, Figure 2). Amygdala, caudate, putamen, midbrain, medulla and cerebellum were significantly different between DLB and AD (p < 0.05).ConclusionDLB‐RAI may aid the differential diagnosis of mild DLB, normal elderly and AD. It may have the potential to be utilized in the clinical practice.