Microstructural grey matter alterations in patients with behavioural variant frontotemporal dementia

TL;DR

Texture analysis of MRI scans can detect early brain changes in behavioral variant frontotemporal dementia that are not visible through traditional volume measurements.

Contribution

The study introduces texture analysis as a novel biomarker for early detection and tracking of behavioral variant frontotemporal dementia.

Findings

Texture analysis detected microstructural changes in mild dementia patients compared to healthy controls.

Texture features identified alterations in moderate dementia even when volume differences were absent.

Texture analysis showed increased sensitivity for tracking disease progression compared to volumetric measures.

Abstract

Behavioural variant frontotemporal dementia is characterized by progressive changes to personality and behaviour, yet early detection and disease staging remain challenging. Current clinical neuroimaging relies on visual assessment of atrophy patterns, which may overlook subtle structural changes. While volumetric analysis has improved the ability to detect neurodegeneration and track disease progression, it may lack the sensitivity to identify microstructural alterations that precede frank atrophy. Texture analysis, a quantitative approach that evaluates the spatial regularity of grey matter density, has demonstrated promise in detecting early microstructural changes in Alzheimer’s disease and distinguishing frontotemporal dementia subtypes. However, its potential as a biomarker for early behavioural variant frontotemporal dementia detection and disease staging remain unexplored. This study evaluates the potential of autocorrelation-based texture features as biomarkers for detecting early-stage behavioural variant frontotemporal dementia and tracking disease progression, comparing their sensitivity and specificity to regional brain volume. We analysed structural MRI scans from behavioural variant frontotemporal dementia patients with mild (n = 21, 61.5 ± 8.5 years, 4.7% female) and moderate dementia (n = 11, 64.4 ± 9.3, 18.1% female) alongside healthy controls (n = 33, 63.1 ± 7.9 years; 36.3% female). Texture and volumetric measures were extracted from frontotemporal regions implicated in behavioural variant frontotemporal dementia pathology. First, we compared these features between healthy controls and patients with mild dementia to identify regions relevant for early diagnosis. Second, we compared patients with mild versus moderate dementia to identify regions linked to disease stage. Analyses were performed both within a composite frontotemporal region of interest and within 160 frontotemporal subregions. We applied false discovery rate correction for multiple comparisons. Microstructural abnormalities captured by texture analysis and volumetric reductions were significantly lower in patients with mild dementia compared to healthy controls within the composite frontotemporal and many subregions (PFDR  < 0.05). In moderate dementia, texture features detected alterations in composite frontotemporal (PFDR  < 0.05) and subregions, including the anterior cingulate, insula and orbitofrontal cortices (PFDR  < 0.05) even when volumetric differences were absent. This study demonstrates that texture-based MRI metrics provide a sensitive measure of microstructural alterations in behavioural variant frontotemporal dementia, detecting some disease-related changes even in regions without detectable volumetric reductions. While volumetric measures are effective for identifying individuals with early-stage disease, texture analysis may offer increased sensitivity for tracking disease progression. Longitudinal studies are needed to validate the predictive value of texture features for clinical decline in behavioural variant frontotemporal dementia.

Akbarian et al. report that texture analysis of MRI detects subtle microstructural grey matter alterations in patients with behavioural variant frontotemporal dementia. These alternations are observed even when volumetric differences are absent, indicating that texture features may serve as a complementary biomarker for early diagnosis and for tracking disease progression.

Graphical Abstract

Figures