Medial temporal lobe atrophy in preclinical dementia: Visual and automated rating during six‐year follow‐up

Abstract

AbstractBackgroundMedial temporal lobe atrophy (MTA), a morphological hallmark of neurodegenerative diseases, especially Alzheimer’s disease, is assessed through visual ratings in the clinic. Here, we study the potential of detecting longitudinal changes in medial temporal atrophy in preclinical dementia with Scheltens' MTA scale (ranging from 0‐4) with ratings from two radiologists and AVRA―our newly developed software providing continuous MTA ratings.MethodIn this study 93 individuals, followed up three times over six years with subjective cognitive decline (SCD, N = 61) or mild cognitive impairment (MCI, N = 32) from the Swedish BioFINDER cohort, were included. Two blinded raters visually assessed MTA according to Scheltens' scale. These ratings were compared to those from AVRA, MMSE, ADAS‐delayed word recall test, as well as hippocampal (HC) and inferior lateral ventricle (ILV) volumes segmented with FreeSurfer.ResultsDifferent rating styles, with one rater being more conservative, resulted in relatively low inter‐rater agreement for visual MTA assessment between radiologists (weighted kappa of 0.33). However, both sets of ratings showed strong Spearman correlations between raters (rs = 0.76) and to AVRA (mean rs = {0.75, 0.84}). Both visual and automated ratings were strongly associated with HC (rs = {−0.55, −0.50, −0.54}, for rater 1, rater 2 and AVRA, respectively) and ILV (rs = {0.79, 0.83, 0.87}) volumes as well as MMSE (rs = {−0.37, −0.34, −0.31}) and ADAS (rs = {0.42, −0.44, 0.40}). AVRA provided average annual progression rates of MTA points for SCD individuals (0.05/year) and MCI patients (0.09/year), with similar rates for the radiologists' ratings. These changes were comparable to the average changes in HC (−2.9%/year and −4.9%/year, for SCD and MCI respectively) and ILV volumes (10.5%/year and 13.0%/year).ConclusionOur results suggest that visual MTA ratings are sensitive to detect longitudinal changes in HC and ILV volumes in SCD and MCI individuals. The main limitation of quantifying atrophy with visual ratings in clinics is the subjectiveness of the assessment.