Exploring the mystery of the human brain is the most cutting-edge and challenging scientific problem in this century. At present, the development and maturity of advanced diagnosis and imaging technology make it possible for us to explore the functional structure of the brain from different angles, including functional magnetic resonance imaging (fMRI), structural MRI, and electroencephalogram (EEG), magnetoencephalogram (MEG), etc. These multimodal data have their own advantages in spatial and temporal resolution, which helps us to better understand the working mechanism of the brain from the perspective of functional integration and explore the psychophysiology mechanism of various brain diseases.
The traditional processing and analysis methods mainly rely on expert systems, which have run into the bottleneck of incomplete knowledge and cannot meet the demand of the big data era. In recent years, deep learning becomes an increasingly reliable option that is capable of finding out the intrinsic, and potentially valuable knowledge from the vast amounts of multimodal brain data, yielding more intelligent data processing and analysis. However, there are still challenges in deep learning-based multimodal brain data analysis. On the one hand, multimodal brain data is quite different from the conventional image and audio data in terms of data dimension and distribution. Designing appropriate deep learning networks is of significance for improving the diagnosis and prognosis. On the other hand, clinical decision-making needs not only high performance but also strong interpretability. It is also worth studying to light up the black-box attribute of deep learning and find potential biomarkers in multimodal brain data.
Original research manuscripts utilizing deep learning methods in multimodal brain data processing and analysis are welcome to contribute to this Research Topic, including but not limited to the following subjects:
• Advanced deep learning approaches for multimodal brain data classification/mapping, including but not limited to structural MRI, functional MRI, diffusion-weighted imaging, MR-spectroscopy, PET, EEG, MEG, CT, etc.
• Innovative multimodal brain data fusion approaches and analysis.
• Lightweight deep networks for a low-level vision of multimodal brain data.
• Neural architecture search for multimodal brain data processing and analysis.
• Novel Multimodal or crossmodal feature learning and retrieval of brain data.
• Interpretable deep learning methods for multimodal brain data analysis and biomarkers discovery.
• Cutting-edge applications and validation on multimodal brain data.
Exploring the mystery of the human brain is the most cutting-edge and challenging scientific problem in this century. At present, the development and maturity of advanced diagnosis and imaging technology make it possible for us to explore the functional structure of the brain from different angles, including functional magnetic resonance imaging (fMRI), structural MRI, and electroencephalogram (EEG), magnetoencephalogram (MEG), etc. These multimodal data have their own advantages in spatial and temporal resolution, which helps us to better understand the working mechanism of the brain from the perspective of functional integration and explore the psychophysiology mechanism of various brain diseases.
The traditional processing and analysis methods mainly rely on expert systems, which have run into the bottleneck of incomplete knowledge and cannot meet the demand of the big data era. In recent years, deep learning becomes an increasingly reliable option that is capable of finding out the intrinsic, and potentially valuable knowledge from the vast amounts of multimodal brain data, yielding more intelligent data processing and analysis. However, there are still challenges in deep learning-based multimodal brain data analysis. On the one hand, multimodal brain data is quite different from the conventional image and audio data in terms of data dimension and distribution. Designing appropriate deep learning networks is of significance for improving the diagnosis and prognosis. On the other hand, clinical decision-making needs not only high performance but also strong interpretability. It is also worth studying to light up the black-box attribute of deep learning and find potential biomarkers in multimodal brain data.
Original research manuscripts utilizing deep learning methods in multimodal brain data processing and analysis are welcome to contribute to this Research Topic, including but not limited to the following subjects:
• Advanced deep learning approaches for multimodal brain data classification/mapping, including but not limited to structural MRI, functional MRI, diffusion-weighted imaging, MR-spectroscopy, PET, EEG, MEG, CT, etc.
• Innovative multimodal brain data fusion approaches and analysis.
• Lightweight deep networks for a low-level vision of multimodal brain data.
• Neural architecture search for multimodal brain data processing and analysis.
• Novel Multimodal or crossmodal feature learning and retrieval of brain data.
• Interpretable deep learning methods for multimodal brain data analysis and biomarkers discovery.
• Cutting-edge applications and validation on multimodal brain data.
