Special Issue on Recent Advances in Deep Learning for Neuroimaging
摘要截稿:
全文截稿: 2018-12-31
影响因子: 1.902
期刊难度:
CCF分类: 无
中科院JCR分区:
• 大类 : 心理学 - 3区
• 小类 : 计算机:人工智能 - 4区
• 小类 : 神经科学 - 4区
• 小类 : 心理学:实验 - 4区
Overview
Regional brain actions and inter regional associations are altered from one mental state to another or from healthy to unhealthy status. This variation might be manifested through neuroimaging signals that are measured from the brain such as electroencephalogram (EEG) has great temporal resolution which explores the transient alterations in brain. However, it is still challenging to extract intrinsic characteristics from these diverse signals because they give more information related to brain structure. In general, automated techniques are applied in the analysis of neuroimaging data including magnetic resonance imaging (MRI) and functional MRI (fMRI). Recently, developments in machine learning methods are combined with computational power of deep learning methods that is proved to be more efficient to solve problems in the analysis of neuroimaging data.The advancements in deep (artificial) neural network models lead towards the robust feature learning methods to attack difficult problems in neuroimaging segmentation and classification. Many impressive outcomes are achieved in the computerized deep learning (DL) methods using imaging databases. DL models commonly need annotations of many images for using supervised learning methods and are one of the roadblocks in using these models in several classification tasks in fMRI/MRI. Unsupervised methods with DL are applied successfully on the classification of natural images. These methods are also used to solve the challenging problems in brain imaging analysis. Domain transfer convolutional neural networks (CNN) are used with end to end DL method and provide better results. An extreme learning machine (ELM) is a type of DL also utilized for neuroimaging analysis. Many automated pipelines with DL methods are applied on nonmedical data for training due to lack of availability of more labeled data. More advancement in the analysis of natural images with DL techniques are applied in NeuroImage and challenges in acquiring the annotations/labels/datasets, adapting/improvising DL models, set up of parameters, multi modality pose generalization are also considered. Although limited results are available in existing work with DL based automated methods for the analysis of NeuroImage, however, it is still believed that the future is brighter in solving hard problems in the analysis of NeuroImage.
Submissions of new research ideas are encouraged in this special issue related to below mentioned topics, but not limited to:
- Analysis of Neuroimaging data based on deep learning architecture
- Differences in representation of normal brain tissues and unhealthy tissues through deep convolutional networks (CNN)
- Brain lesion detection using deep CNN models with exploration of feature learning
- Decoding mental stateusing deep CNNarchitecture
- Unsupervised learning with deep learning methods for brain diseases detection
- Extreme learning and Boltzmann machine for NeuroImage analysis
- Long short term memory model (LSTM) using brain diseases detection
- Supervised learning with deep learning methods for brain imaging analysis
- Multi transfer learning with CNN methods for the analysis of brain patterns
- Comprehensive review on brain lesion segmentation and classification
- 3D modeling of brain structure using CNN architecture
- Cardiovascular disease detection using deep learning
- Novel machine learning techniques based on neuroglial brain network
- Deep learning optimizationn in neuroscience
- Review of machine learning methods to solve real life neuroscience problems
- Brain age estimation using Deep learning techniques from MRI