Precise perivascular space segmentation on magnetic resonance imaging from Human Connectome Project-Aging.

Area of Science:

• Neuroimaging
• Cerebrovascular Health
• Aging Research

Background:

• Perivascular spaces (PVS) are fluid-filled tunnels around brain vasculature, integral to the glymphatic system.
• Alterations in PVS are associated with aging and cerebrovascular diseases, highlighting the need for accurate segmentation.
• Current PVS segmentation methods face challenges due to small structure size, variable MRI appearance, and limited annotated data.

Purpose of the Study:

• To present a high-quality, manually corrected dataset of segmented perivascular spaces (PVS) from T2-weighted MRI scans.
• To facilitate research into PVS changes across the adult lifespan (ages 30-100) and their association with cognitive decline.
• To support the development of advanced automated medical image segmentation algorithms for neuroimaging.

Main Methods:

• Utilized T2-weighted MRI scans from 200 subjects (ages 30-100) from the Human Connectome Project Aging (HCP-Aging) cohort.
• Employed a hybrid approach combining unsupervised and deep learning techniques for initial PVS segmentation.
• Incorporated manual corrections to ensure high accuracy and reliability of the segmented PVS dataset.

Main Results:

• Generated a comprehensive dataset of finely segmented perivascular spaces (PVS).
• The dataset covers a wide age range, enabling age-related PVS analysis.
• The methodology ensures high accuracy, suitable for detailed scientific investigation.

Conclusions:

• The presented PVS dataset is a valuable resource for studying PVS dynamics in aging and neurodegeneration.
• This dataset will advance the development of automated segmentation tools for improved medical imaging analysis.
• Facilitates research linking PVS changes to cognitive function and early detection of neurological disorders.