Brain Image Analysis And Calculation Based On Deep Neural Network

Human brain plays a vital role in human life activities,which has complex structure and various functions.Imaging technology with advantages of its painless and non-invasive specialty,has been widely used in clinical practice.Using brain imaging to explore the mysteries of brain structure and function is a powerful means for studying brain.With the development of neuroimaging,and a large amount of data accumulated.Computer-assisted image analysis can reduce the workload of doctors and increase the efficiency for disease diagnosis,is of great potential in near future.The existing methods of brain imaging analysis can be divided into five steps,which are image pre-processing,image registration,image segmentation,feature extraction and classification.The whole procedure is complicated and the performance is heavily dependent on handcraft feature engineering.Motivated by the success of deep learning in image classification,this thesis studies deep neural network algorithm for brain imaging diagnosis,which can extract feature automatically through training neural network via data.This thesis uses magnetic resonance imaging and positron emission tomography imaging for the purpose of distinguishing Alzheimer's disease and mild cognitive impairment.No image segmentation and rigid registration are required in our procedure.This thesis has two main contributions:(1)This thesis proposes a new classification framework based on combination of 2D convolutional neural networks(CNN)and recurrent neural networks(RNN),which learns the features of 3D PET images by decomposing the 3D image into a sequence of 2D slices.In this framework,the hierarchical 2D CNNs are built to capture the intra-slice features while the gated recurrent unit(GRU)of RNN is used to extract the inter-slice features for final classification.Our method is evaluated on the baseline PET images from 339 subjects including 93 AD patients,146 mild cognitive impairments(MCI)and 100 normal controls(NC)from Alzheimer's Disease Neuroimaging Initiative(ADNI)database.Experimental results show that the proposed method achieves an accuracy of 91.19% for classification of AD vs.NC and 77.96% for classification of MCI vs.NC,respectively,demonstrating the promising classification performance.(2)This thesis proposes to construct cascaded convolutional neural networks(CNNs)to learn the multi-level and multimodal features of MRI and PET brain images for AD classification.First,multiple deep 3D-CNNs are constructed on different local image patches to transform the local brain image into more compact high-level features.Then,an upper high-level 2DCNN followed by softmax layer is cascaded to ensemble the high-level features learned from the multi-modality and generate the latent multimodal correlation features of the corresponding image patches for classification task.Finally,these learned features are combined by fully connection followed by softmax layers for AD classification.The proposed method can automatically learn the generic multi-level and multimodal features from multiple imaging modalities for classification,which are robust to the scale and rotation variations to some extent.Experimental results show that the proposed method achieves an accuracy of 93.26% for classification of AD vs.NC and 82.95% for classification pMCI vs.NC,demonstrating the effectiveness of proposed method.