Quality Assessment of Brain MRI Defacing Using Machine Learning.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Data Privacy

Background:

• Defacing brain magnetic resonance imaging (MRI) scans is vital for patient privacy in research.
• Current defacing methods have errors, risking patient anonymity.
• Automated quality assessment is needed to ensure accurate defacing.

Purpose of the Study:

• To investigate the feasibility of automated quality assessment for defaced brain MRIs.
• To evaluate the efficacy of machine learning (ML) models in this task.

Main Methods:

• Utilized machine learning models for automated quality assessment.
• Trained models to distinguish between properly and inadequately defaced MRI scans.

Main Results:

• Machine learning models demonstrated high accuracy in identifying defacing errors.
• The proposed ML approach shows promise for reliable quality assessment.

Conclusions:

• Automated quality assessment using ML is feasible and effective for defaced brain MRIs.
• ML offers a robust solution for maintaining data integrity and patient anonymity in medical imaging.