A method to improve interrater reliability of visual inspection of brain MRI scans in dementia

Abstract

MR scanning is used in the clinical evaluation of patients with dementia but lacks a reliable method of visual inspection. Two neurologists conducted multiple pilot trials of alternate methods for visual inspection of MRIs, including methods that produced at least 75% interrater agreement in repeat trials, and selected a final method for rating ventricular:brain ratio (VBR), cortical atrophy, and white matter changes. Two other neurologists, new to the method, tested interrater reliability for each component of the method after a brief training session. The correlation of VBR measurement was 0.884 (p = 0.0001). The weighted kappa scores were 0.68 for overall frontal lobe atrophy, 0.38 for right temporal lobe atrophy, 0.20 for left temporal lobe atrophy, and 0.54 for parietal lobe atrophy. The weighted kappa scores were 0.77 for overall periventricular white matter hyperintensities and 0.72 for centrum semiovale hyperintensities. The proposed method may provide a rapid and reliable way to assess VBR, frontal lobe atrophy, parietal lobe atrophy, and white matter changes on brain MRIs in the evaluation of dementia, but it was less reliable for the assessment of temporal lobe atrophy.

Dilated Virchow-Robin Spaces in Myotonic Dystrophy: Frequency, Extent and Significance

“…2. Three-step scales for rating methods, which are based on reference to standard images, have produced the better interrater agreement [23]. …”

“…The width of the ventricular bodies halfway from the anterior to posterior margin was divided by the maximum width of the brain at that level. VBR measurements have shown excellent interrater reliability [23]. …”

Brain Magnetic Resonance Imaging White Matter Lesions Are Frequent in HTLV-I Carriers and Do Not Discriminate from HAM/TSP

“…Small lesions were 3-6 mm, large lesions >6 mm in largest dimension, and confluent lesions were irregular in shape and >6 mm. Brain atrophy was quantified by visual inspection as described by Victoroff et al 26 Briefly, three brain regions (frontal, parietal, and temporal lobes) were rated from zero (no atrophy) to three (severe atrophy) and the total score was the sum of the three regions (possible score range zero to nine). Total lesion volume was calculated from the summed cross-sectional area of FLAIR hyperintense lesions using ImageJ software.…”

Visual Rating versus Volumetry to Detect Frontotemporal Dementia

“…Neuroradiologists primed to the diagnosis of behavioral variant FTD versus semantic dementia versus AD show higher accuracy of diagnostic categorization and inter-rater agreement than the neurologists in this and a previous study [27], although sensitivity remains low at approx. 60% [33].…”

“…Our experienced clinicians could not differentiate controls from FTD at the 90% accuracy achieved by semi-automated volumetry [21]. This has been a challenge for several investigators [24,25,26,27]. This difficulty echoes the conclusion of our prior paper showing that many frontotemporal regions are mildly atrophied with normal aging and may overlap with FTD-related mild atrophy [14].…”