Semi-Supervised Learning in Medical Image Segmentation: Brain Tissue with White Matter Hyperintensity Segmentation Using FLAIR Image

White matter hyperintensity (WMH) has been considered the primary biomarker from small-vessel cerebrovascular disease to Alzheimer’s disease (AD) and has been reported for its correlation of brain structural changes. To perform WMH related analysis with brain structure, both T1-weighted (T1w) and (Fluid Attenuated Inversion Recovery(FLAIR) are required. However, in a clinical situation, it is limited to obtain 3D T1w and FLAIR images simultaneously. Also, the most of brain segmentation technique supports 3D T1w only. Therefore, we introduced the semi-supervised learning method that can perform brain segmentation using FLAIR image only. Our method achieved a dice overlap score of 0.86 for brain tissue segmentation on FLAIR, with the relative volume difference between T1w and FLAIR segmentation under 4.8%, which is just as reliable as the segmentation done by its paired T1w image. We believe our semi-supervised learning method has a great potential to be used to other MRI sequences and provide encouragement to people who seek brain tissue segmentation from a non-T1w image.

Direct clipping of large basilar trunk aneurysm

A large basilar trunk aneurysm was incidentally found in a 77-year-old woman in examination for headache. Though it was asymptomatic, high signal intensity was noticed in the brainstem around the aneurysm on FLAIR image of MRI. As she was otherwise healthy, surgical clipping was performed through anterior temporal approach.The video can be found here: http://youtu.be/0soWM8meCW8.

Abstract WP122: The Coexisting Leukoaraiosis In Ischemic Stroke Patients: How Does It Change Over Time?

Background: The prevalence of leukoaraiosis or white matter hyperintensities (WMH) is getting higher with the easy accessibility of brain imaging. Their clinical significance has not been known yet. Herein, we aimed to investigate how they progress or regress over time in patients with ischemic stroke. Methods: We consecutively enrolled ischemic stroke patients who underwent brain MRI (T2-weighted image or fluid-attenuated inverse recovery (FLAIR) image) more than twice with at least a 6 month time-interval. Patients with pathologic lesions which cannot be distinguished from WMH were excluded. Clinical variables (age, sex, risk factors, stroke subtypes, and imaging time) were obtained by reviewing medical records. T2-weighted image or FLAIR image was used to assess the presence of WMH. WMH was visually measured with use of modified Fazekas grading by an independent investigator. The change of WMH was interpreted as 1)no change, 2)progress, 3)regression, 4)distinctive progress and regression. The descriptive analysis of the pattern of WMH change over time and factors associated with each pattern were analyzed. Results: The total number of enrolled patients were 100. Their age (mean±SD) was 67.5±11.8 years and 63 were male. The most common stroke subtype was small vessel disease (SVD, 43%), followed by large vessel disease (LVD, 27%), and cardioembolism (9%). The imaging time-interval (mean±SD) was 28.0±16.6 months. In 64 patients, WMH was observed on brain MRI. Among them, the WMH pattern didn’t change in 25 patients, progressed in 28, regressed in 10 and progressed in distinctive region and regressed in others in 4 patients. Three patients had newly developed WMH on their follow-up MRI. 51.9% (14/27) of patients with LVD showed progress, whereas 30.2% (13/43) of SVD showed progress. Regarding regression, 16.3% (7/43) of patients with SVD, 11.1% (3/27) of LVD showed regression. The presence of LVD was significantly associated with WMH progression (14/27 vs. 18/73, p=0.015). The WMH changing pattern was not affected by risk factors. Conclusions: Over half of patients with stroke showed progress of WMH on their follow-up MRI. In addition, regression of WMH was also observed in 21.8% patients. Stroke subtype of LVD was associated with WMH progression.