[论文精读]Informative Biomarkers for Autism Spectrum Disorder Diagnosis in Functional Magnetic Resonance

论文全名：Informative Biomarkers for Autism Spectrum Disorder Diagnosis in Functional Magnetic Resonance Imaging Data on the Default Mode Network

论文网址：Applied Sciences | Free Full-Text | Informative Biomarkers for Autism Spectrum Disorder Diagnosis in Functional Magnetic Resonance Imaging Data on the Default Mode Network (mdpi.com)

英文是纯手打的！论文原文的summarizing and paraphrasing。可能会出现难以避免的拼写错误和语法错误，若有发现欢迎评论指正！文章偏向于笔记，谨慎食用

目录

1. 省流版

1.1. 心得

1.2. 论文总结图

2. 论文逐段精读

2.1. Abstract

2.2. Introduction

2.3. Materials and Methods

2.3.1. fMRI Data Acquisition

2.3.2. Inclusion/Exclusion Criteria and Pre-Processing

2.3.3. Feature Extraction

2.3.4. Feature Selection and Biomarker Assessment

2.3.5. Classification

2.4. Results

2.4.1. Classification Performance Results

2.4.2. Analysis of Selected Biomarkers

2.4.3. Functional Connectivity Features Validation

2.4.4. Evaluation of the Adopted Framework and Relationship to Prior Work

2.5. Discussion

2.5.1. Classification Performance

2.5.2. Informative Biomarkers

2.6. Conclusions

3. 知识补充

3.1. recursive feature elimination with correlation bias reduction (RFE-CBR)

4. Reference

---

1. 省流版

1.1. 心得

（1）四区的文章。但完全不是一篇烂文章，表述和方法都还可以，顶多只是方法没太有创新罢了

（2）⭐将头动作为参数！！！

1.2. 论文总结图

2. 论文逐段精读

2.1. Abstract

        ①Different sites or individuals may distrub the finding of biomarkers.

        ②Task: classify TD and ASD and identify the biomarkers

potent  adj. 强有力的，有影响力的；（议论等）有说服力的；浓烈的；（药品等）有强效的；（男子）有性能力的

2.2. Introduction

        ①Introduce ASD→FC→AI with model accuracy examples→model performances with different atlases→their methods

        ②There might be difference in DMN between TD and ASD

substrate  n.基底；底物；底层；基层

2.3. Materials and Methods

2.3.1. fMRI Data Acquisition

        ①Dataset: ABIDE I

        ②Samples: total 1112 participants with 539 ASD and 573 TD（后面还要筛选）

2.3.2. Inclusion/Exclusion Criteria and Pre-Processing

        ①Sample filtering: excluding which movement higher than 0.2 mm, then 883 left; 进一步检查了构成DMN的18个大脑区域的BOLD时间序列，删除了其中一个 BOLD 时间序列为零的个体数据。Finally left 871 participants with the portion between ASD and TD individuals is approximately 1.18

2.3.3. Feature Extraction

        ①They include demographic information and patient movement information

（1）Static Functional Connectivity

        ①Atlas: CC200

        ②Selected brain regions: 153 from 18 DMN regions

        ③sFC: calculated by Pearson using pairs of BOLD time series:

（2）Dynamic Functional Connectivity

        ①dFC: calculated by window slicing with 65 length, 移动一个重复时间（TR）点，该点可以是几个时间点，具体取决于聚合采集时间。153 features as well.

（3）Demographics

        ①Applied 2 demographic data: age and handedness（那为什么要把性别放进来？？）（根据后面特征数量的计算，这里确实没有加入性别）

ambidextrous  adj.左右手都灵巧的；左右开弓的

（4）Acquisition Parameters

        ①9 parameters: "field of view (FOV, 3 features—one for each dimension, x, y, and z), repetition time (TR, 1 feature), echo time (TE, 1 feature), voxel size (3 features corresponding to space dimensions), and number of slices (1 feature)"

（5）Motion Parameters

        ①⭐Head motion parameter: the parameters are calculated by "framewise displacement (FD) provided from each datacenter (2 features corresponding to FD mean value and percentage of slices with FD over 0.2 mm)"

        ②Rate of change in the BOLD signal (DVARS): calculated by "temporal derivative of time courses and the variance over voxels of root mean square"

2.3.4. Feature Selection and Biomarker Assessment

        ①Workflow of proposed method:

        ②⭐作者这里说有473（153 of sFC+306 of dFC+2 of demographics+9 of acqusition+3 of motion）个特征/人，但是一共有871个人。作者引用(Ambroise & McLachlan, 2002)说明特征参数最好是被试的三分之一不然容易过拟合（但是有点早了，也不能说完全可靠吧），因此需要引入如下的特征选择方式：

        ③Feature selection (FS) method: recursive feature elimination with correlation bias reduction (RFE-CBR)（类似SVM每次删除最不重要的以及除去高度相关的特征）

        ④Cross validation: 10 fold

2.3.5. Classification

        ①Classifier: SVM

        ②Optimize: minimizing the 

        ③Set ratio: 90% for training and 10% for testing 

        ④Maximum epoch: 50

2.4. Results

2.4.1. Classification Performance Results

        ①Performance: 76.63% ACC, 78.63% SEN, 74.27% SPE and 82.74% AUC with 136 features（不知道为什么ABIDE数据集的AUC都普遍比ACC高）

        ②⭐作者觉得ASD患者性别分布悬殊所以性别也是潜在的生物标志物

        ③AUC figure:

        ④136 feature included (dFC most):

included	a) dFC characteristics (47 and 45 out of the 136 total features for mean dFC and variance dFC, respectively), b) sFC incorporating 43 (out of the 136) c) acqusition: echo time (TE)
excluded	demographics, motion, and the rest of the acquisition features (FOV, voxel size, and repetition time)

        ⑤Discriminative features:

where green line denotes sFC, blue denotes variance dFC, red denotes mean value dFC

2.4.2. Analysis of Selected Biomarkers

        ①Lists discriminative brain regions

2.4.3. Functional Connectivity Features Validation

        ①They employed Wilcoxon test on ASD and TD, finding that there are significant differences between 19 features. These significant differences includesFC and mean dFC

2.4.4. Evaluation of the Adopted Framework and Relationship to Prior Work

        ①Performance comparison table:

2.5. Discussion

2.5.1. Classification Performance

        ①作者觉得不同的数据选择让方法之间无法比较，我的评价是得自己去复现别人的代码在自己选的数据上。这里作者确实可能没去做这个工作量

        ②作者觉得必须筛选一些不好的数据，其实这个还是备受争议啦，毕竟人家ABIDE采的时候就筛选过了

2.5.2. Informative Biomarkers

        ①Introducing some medical speculation on ASD

plethora  n.过多；过剩；过量    excitatory  adj.兴奋的；有刺激性的，兴奋的

oscillation  n.振动；摆动；振幅；摇摆；(情感、行为、思想的)摇摆不定，变化无常，犹豫不定；浮动；一次波动

2.6. Conclusions

        Their model got the most highest performance

3. 知识补充

3.1. recursive feature elimination with correlation bias reduction (RFE-CBR)

（1）论文：https://www.sciencedirect.com/science/article/pii/S0925400515001872

（2）参考学习：【数据处理系列】深入理解递归特征消除法（RFE）：基于Python的应用-CSDN博客

4. Reference

Ambroise, C. & McLachlan, G.J. (2002) ‘Selection bias in gene extraction on the basis of microarray gene-expression data’, Proc. Natl. Acad. Sci. 99: 6562–6566. doi: https://doi.org/10.1073/pnas.102102699

Karampasi, A. S. et al. (2024) 'Informative Biomarkers for Autism Spectrum Disorder Diagnosis in Functional Magnetic Resonance Imaging Data on the Default Mode Network', Appl. Sci., 11. doi: Applied Sciences | Free Full-Text | Informative Biomarkers for Autism Spectrum Disorder Diagnosis in Functional Magnetic Resonance Imaging Data on the Default Mode Network