DeepPrep: an accelerated, scalable and robust pipeline for neuroimaging preprocessing empowered by deep learning.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Artificial Intelligence in Medicine

Background:

• Neuroimaging research is generating big data, posing significant computational challenges.
• Existing preprocessing pipelines struggle to keep pace with the expanding volume of neuroimaging data.
• Scalability and efficiency are critical for modern neuroimaging research.

Purpose of the Study:

• To introduce DeepPrep, a novel pipeline designed to overcome computational bottlenecks in neuroimaging.
• To leverage deep learning and workflow management for accelerated data preprocessing.
• To evaluate the performance of DeepPrep against state-of-the-art methods.

Main Methods:

• Development of DeepPrep, a pipeline integrating deep learning algorithms.
• Implementation of a workflow manager to enhance processing efficiency.
• Large-scale evaluation using over 55,000 neuroimaging scans.

Main Results:

• DeepPrep achieved a tenfold acceleration in processing speed compared to existing pipelines.
• Demonstrated significant improvements in scalability to handle large datasets.
• Confirmed robustness of the DeepPrep pipeline across diverse neuroimaging data.

Conclusions:

• DeepPrep effectively addresses the computational challenges of big data in neuroimaging.
• The pipeline offers a scalable and robust solution for accelerated preprocessing.
• DeepPrep meets the demanding scalability requirements of contemporary neuroimaging research.