Stroke subtype classification by geometrical descriptors of lesion shape

Area of Science:

• Neurology
• Radiology
• Medical Imaging

Background:

• Inferring stroke etiology from acute MRI lesion patterns aids patient management and prognosis.
• Classifying stroke subtypes, particularly cardioembolic stroke, is clinically significant.

Purpose of the Study:

• To evaluate the utility of 3D geometrical lesion-shape descriptors for classifying stroke subtypes.
• To specifically assess the ability of these descriptors to identify cardioembolic stroke origins.

Main Methods:

• Retrospective analysis of acute stroke patients classified by ASCOD etiology.
• Segmentation of lesions on diffusion-weighted MRI datasets.
• Quantification of lesion shape descriptors including surface area, sphericity, and volume ratios.

Main Results:

• Small vessel disease lesions were smaller and more spherical than other stroke subtypes.
• No significant morphological differences were found between cardioembolic and non-cardioembolic stroke groups.
• 150 patients were analyzed (mean age 77 years).

Conclusions:

• Distinct morphological properties of stroke lesions can differentiate small vessel disease from other stroke types.
• Lesion morphology, as analyzed, was insufficient to infer the etiology of embolic strokes.