ABSTRACT
BACKGROUND: Medial lemniscus T2 hyperintensity (MLH) has been recently demonstrated as potential imaging marker for small vessel disease (SVD). Our purpose in this study is to improve accuracy of regions of interest (ROI) analysis for this imaging finding. METHODS AND METHODS: Two neuroradiologists retrospectively reviewed 103 consecutive outpatient brain MRI. Medial lemniscus signal in dorsal pons was evaluated; visually on FLAIR and with ROI on T2. Original MRI interpretations were divided into three categories; SVD, multiple sclerosis (MS), and nonspecific WM changes (non). RESULTS: Thirty-seven patients had SVD, 14 patients had MS, 52 had Non. Visual MLH was seen exclusively with SVD and was generally bilateral. Patients with visual MLH belonged to advanced SVD by imaging and clinical parameters. Compared to visual data, ROI analyses of MLH has been known to be compounded by false positives and negatives at low threshold (20% of adjacent to normal brainstem signal). With application of higher ROI threshold (25%), false positives were eliminated but false negatives increased. ROI analyses of MLH by experienced neuroradiologist were more reliable. CONCLUSION: MLH seen on high threshold ROI analysis is a reliable radiologic marker in predicting SVD. ROI analysis of MLH should be performed by an experienced neuroradiologist.
Subject(s)
Cerebrovascular Disorders/pathology , Image Enhancement/methods , Magnetic Resonance Imaging/methods , Medulla Oblongata/blood supply , Medulla Oblongata/pathology , Multiple Sclerosis/pathology , White Matter/pathology , Aged , Algorithms , Diagnosis, Differential , Female , Humans , Image Interpretation, Computer-Assisted/methods , Male , Microvessels/pathology , Middle Aged , Reproducibility of Results , Sensitivity and SpecificityABSTRACT
OBJECTIVE: Small-vessel disease is a common MRI finding that can be difficult to differentiate from other white matter (WM) diseases because of the lack of a specific pattern of brain involvement. The purpose of our study was to evaluate medial lemniscus hyperintensity seen on FLAIR images as an imaging marker for small-vessel disease. MATERIALS AND METHODS: Two blinded neuroradiologists retrospectively reviewed 103 consecutive outpatient brain MRI studies. Medial lemniscus signal in the dorsal pons was evaluated visually on FLAIR images and after placing regions of interest (ROIs) on T2-weighted images. On the basis of the original interpretations, scans were divided into three categories: small-vessel disease, multiple sclerosis (MS), and normal or nonspecific WM changes. Cardiovascular risk factors were recorded. Analysis of variance and Fisher exact tests were used to determine group differences, and kappa statistics was used to determine interrater agreement. RESULTS: Thirty-seven patients had small-vessel disease, 14 patients had MS, and 52 had nonspecific WM changes. Medial lemniscus hyperintensity was seen in about 20% of patients with small-vessel disease and was generally bilateral. Although ROI analyses identified a slightly higher number of patients with medial lemniscus signal > 20% of adjacent to normal-appearing brainstem, interrater reliability was moderate, and there were false-positive and false-negative cases in comparison with visual data. When small-vessel disease patients were further subdivided into mild or advanced subgroups, medial lemniscus hyperintensity was selectively seen in advanced small-vessel disease. Patients with medial lemniscus hyperintensity were older (p < 0.001) and had higher prevalence of diabetes (p = 0.03), hypertension (p = 0.009), and hypercholesterolemia (p = 0.03). CONCLUSION: Medial lemniscus hyperintensity seen on FLAIR images is a reliable radiologic marker of advanced small-vessel disease.
