Enhanced brain extraction improves the accuracy of brain atrophy estimation

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Brain Extraction Tool (BET) is crucial for separating brain from non-brain tissues in MRI.
• SIENA and SIENAX quantify brain volume and atrophy using BET, but occasional inaccuracies require manual correction.
• Manual editing is time-consuming and introduces user variability.

Purpose of the Study:

• To introduce and evaluate a new, fully automated BET option for improved brain extraction.
• To compare the performance of the new automated BET with traditional BET and manual editing procedures.
• To assess the impact of the new BET on brain volume and atrophy measurements in SIENA and SIENAX.

Main Methods:

• Developed a new automated BET option incorporating standard-space masking and morphological operations for refined tissue removal.
• Compared brain volume and atrophy measures obtained using the traditional BET, manual editing, and the new automated BET within SIENA and SIENAX.
• Evaluated the overlap of excluded voxels between manual editing and the new automated BET using the Dice overlap coefficient.

Main Results:

• The new automated BET procedure yielded brain volume and atrophy measures nearly identical to manual editing.
• Both manual editing and the new automated BET provided significantly different (more accurate) atrophy measures compared to the traditional BET.
• High correlation and Dice overlap coefficients confirmed the similarity between manual editing and the new automated BET outcomes.

Conclusions:

• The proposed automated BET procedure achieves results comparable to manual editing, offering a more accurate estimation of true brain volume.
• This unbiased, fully automated method can benefit multicenter studies monitoring brain atrophy in clinical trials.
• The new BET option reduces the need for manual intervention, saving time and improving consistency in neuroimaging analysis.