Automated Lesion and Feature Extraction Pipeline for Brain MRIs with Interpretability

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Quantitative analysis of brain MRIs is crucial for diagnosing and monitoring neurological conditions.
• Existing tools for lesion segmentation and feature extraction can be complex and lack flexibility.
• There is a need for automated, customizable pipelines that integrate with clinical workflows.

Purpose of the Study:

• To introduce the Automated Lesion and Feature Extraction (ALFE) pipeline, an open-source tool for brain MRI analysis.
• To demonstrate ALFE's ability to perform automated anatomical and lesion segmentation.
• To highlight ALFE's capability in extracting human-interpretable imaging features for clinical and machine learning applications.

Main Methods:

• Development of an open-source, Python-based pipeline named ALFE.
• Implementation of a decoupled design allowing customization of image processing, registration, and AI segmentation modules.
• Modeling the pipeline after established neuroradiology workflows.

Main Results:

• ALFE successfully generates accurate anatomical and lesion segmentations from brain MR images.
• The pipeline extracts quantitative, human-interpretable imaging features describing brain lesions.
• Case studies demonstrate the pipeline's utility in real-world scenarios.

Conclusions:

• The ALFE pipeline offers a flexible and automated solution for brain MRI analysis.
• ALFE facilitates quantitative analysis and machine learning applications by providing standardized lesion features.
• This open-source tool has the potential to advance clinical research and practice in neuroradiology.