Fast semi-automated lesion demarcation in stroke

Area of Science:

• Neuroimaging
• Neurology
• Medical Image Analysis

Background:

• Lesion-behavior mapping requires precise brain lesion demarcation on patient imaging data.
• Manual lesion demarcation is time-consuming and prone to observer variability.
• Existing automated methods are often unsuitable for acute stroke research due to varying imaging modalities.

Purpose of the Study:

• To evaluate the Clusterize algorithm, a semi-automated approach, for lesion demarcation in acute stroke patients.
• To assess the precision, reproducibility, and time efficiency of Clusterize compared to manual methods.
• To determine the algorithm's utility across different common imaging modalities.

Main Methods:

• The Clusterize algorithm was applied to acute stroke patient datasets (CT, DWI, T2FLAIR) and a chronic dataset (T1).
• Performance was compared against manual lesion demarcation standards.
• Processing time, precision, and reproducibility were key evaluation metrics.

Main Results:

• Clusterize significantly reduced lesion demarcation time by an average of 17.8 minutes per patient.
• Time savings increased with larger lesion volumes, reaching up to 60 minutes.
• The algorithm demonstrated comparable precision and reproducibility to manual methods across modalities.
• Performance on chronic T1 images was similar to acute datasets.

Conclusions:

• The semi-automated Clusterize algorithm offers a precise, reproducible, and significantly faster alternative to manual lesion demarcation.
• It is well-suited for acute stroke research with diverse imaging modalities.
• Clusterize, integrated into an SPM toolbox, is recommended for efficient lesion-behavior mapping data preparation.