Classification of Alzheimer’s Disease Using Whole Brain Hierarchical Network
来源:互联网 发布:js中prompt 编辑:程序博客网 时间:2024/05/29 18:11
The main contributions of this paper can be summarized as follows:
A hierarchical network is constructed to provide the complementary information for classification.
New ROIS combinations and feature selection methods are also proposed.
The MKBoost based classification method is used for AD and MCI diagnosis based on MRI data.
1.DATA
The cohort consisted of 200 patients with Alzheimer’s Disease (AD), 120 subjects that had mild cognitive impairment and converted to AD within 18 months (MCIc), 160 subjects with mild cognitive impairment that did not convert (MCInc) and 230 healthy controls (HC).
2. DATA PREPROCESSING
Firstly, non-parametric non-uniform bias correction (N3) algorithm is used to correct intensity inhomogeneity. Secondly, the skull and cerebellum are removed by using BET . Thirdly, each MRI brain image is further segmented into three tissue types, namely gray matter (GM), white matter (WM) and CSF by using FAST . Finally, all brain images are affine aligned by FLIRT .
3.Hierarchical Network Construction
In this article, we define the layers Lc,where c=1,2,3,4, to represent a brain of different number of ROIs as shown in Fig.3. The details of each layer are as follows:
The layer L4 contains 90 original ROIs based on AAL atlas.
The layer L3 contains 54 ROIs. A group of ROIs in L4 are merged into an ROI in L3 if their first two digits and the last digit are the same, for example, Cingulum Ant L (4001), Cingulum Mid L (4011) and Cingulum Post L (4021) are considered as an ROI in the layer L3.
Similar to the process of the layer L3. A group of ROIs in L4 are merged into an ROI in L2 if their first and the last digit are the same, for example,Cingulum Ant L (4001), Cingulum Mid L (4011), Cingulum Post L (4021), Hippocampus L (4101), ParaHippocampal L (4111) and Amygdala L (4201)are considered as an ROI in the layer L2. The layer L2 contains 14 ROIs.
The layer L1 contains only one ROI, that is the whole brain.
4.The Representation and Connectivity of the ROIs
In this section, GLCM is used to extract 3D texture parameters of each ROI indirections.
We then calculate the average texture parameters of each ROI in four directions.
Afterwards, the connectivity between each pair of ROIs is evaluated by Pearsons correlation coefficients.
In this study, the F-score method is employed for feature ranking.
5. Classification and Multiple Kernel Learning
In this paper, we use MKBoost algorithm proposed by Hao et al. [32] for classification tasks.
For each data set,we create a set of 13 base kernels, i.e.,
6. RESULTS AND DISCUSSION
In this article, a 10-fold cross validation strategy is employed to evaluate the performance of the classifier.
Accuracy (ACC), sensitivity (SEN), and specificity (SPE) are used to quantify the performance of the classifier based on the results of 10-fold cross validation.
where TP, FP, TN, and FN are the numbers of true positives,false positives, true negatives, and false negatives, respectively.
In addition, we also report the area under receiver operating characteristic (ROC) curve (AUC). The AUC value is an important index to measure the overall-performance of classification methods, and the higher the AUC value, the better the performance of the classification method.
- Classification of Alzheimer’s Disease Using Whole Brain Hierarchical Network
- 2017-trans-Alzheimer’s Disease Classification 3-D Texture Features
- Hierarchical Attention Network for Document Classification阅读笔记
- Hierarchical Attention Network for Document Classification--tensorflow实现篇
- HD-CNN: HIERARCHICAL DEEP CONVOLUTIONAL NEURAL NETWORK FOR IMAGE CLASSIFICATION(泛读)
- Hierarchical Attention Network for Document Classification中文文本分类,以及模型部署
- Vehicle Type Classification Using Semi-Supervised Convolutional Neural Network(泛读)
- Vehicle Type Classification Using Unsupervised Convolutional Neural Network(泛读)
- Context-Sensitive Twitter Sentiment Classification Using Neural Network
- Implementation of Hierarchical Attention Networks for Document Classification的讲解与Tensorflow实现
- C2. Brain Network (medium)
- Hierarchical task network&STRIPS
- Spam classification using Spark’s DataFrames, ML and Zeppelin
- Robust Classification of Rare Queries Using Web Knowledge
- A. Brain's Photos
- Codeforces C1. Brain Network (easy)
- Codeforces C2. Brain Network (medium)
- 分层匹配 image classification Hierarchical matching
- SAMBA配置 “你可能没有权限访问网络资源”的问题解决方法
- javascript禁止手机浏览器、微信浏览器的上下滑动,滚动问题
- 网络须知UDP vs TCP
- C#文件与数据流
- java根据ip地址获取详细地域信息
- Classification of Alzheimer’s Disease Using Whole Brain Hierarchical Network
- 欢迎使用CSDN-markdown编辑器
- poj 2386 Lake Counting
- 链式前向星
- Linux下修改Mysql的用户(root)的密码
- c++考试总结(二)
- windows 下使用make命令,编译代码
- 迭代器实现Vector
- python的list类型常用操作