




摘要:功能磁共振成像技術(shù)(fMRI: functional Magnetic Resonance Imaging)是一種高效的腦成像技術(shù)研究方法, 為減少fMRI數(shù)據(jù)的冗余, 將其轉(zhuǎn)換為更具分類潛力的特征, 提出一個(gè)基于孿生網(wǎng)絡(luò)(SANet: Siamese Network)的特征構(gòu)造算法SANet, 將多個(gè)掃描點(diǎn)下的腦區(qū)信息類比為圖, 應(yīng)用改進(jìn)的AlexNet網(wǎng)絡(luò)進(jìn)行特征構(gòu)造, 并結(jié)合增量特征選擇策略達(dá)到優(yōu)化分類的目的。通過實(shí)驗(yàn)對(duì)比3種不同網(wǎng)絡(luò)結(jié)構(gòu)和4種分類器對(duì)SANet模型的影響, 并進(jìn)行消融實(shí)驗(yàn), 驗(yàn)證增量特征選擇算法對(duì)SANet構(gòu)造特征的分類效果。實(shí)驗(yàn)表明, SANet模型能對(duì)fMRI數(shù)據(jù)進(jìn)行有效構(gòu)造, 且提高原始特征的分類性能。
關(guān)鍵詞:功能磁共振成像; 特征構(gòu)造; SANet模型; 孿生網(wǎng)絡(luò); 增量特征選擇
中圖分類號(hào): TP399 文獻(xiàn)標(biāo)志碼: A
Siamese Network Based Feature Engineering Algorithmfor Encephalopathy fMRI Images
ZHOU Fengfeng, WANG Qian, DONG Guangyu
(College of Computer Science and Technology, Jilin University, Changchun 130012, China)
Abstract:fMRI(functional Magnetic Resonance imaging) is an efficient research method for brain imaging technique.In order to reduce the redundancy of the fMRI data and transform the fMRI data to the constructed features with more classification potential, a feature construction method based on the siamese network named as SANet(Siamese Network) is proposed. It engineered the brain regions features under multiple scanning points of an fMRI image. The improved AlexNet is used for feature engineering, and the incremental feature selection strategy is used to find the best feature subset for the encephalopathy prediction task. The effects of three different network structures and four classifiers on the SANet model are evaluated for their prediction efficiencies, and the ablation experiment is conducted to verify the classification effect of the incremental feature selection algorithm on the SANet features. The experimental data shows that the SANet model can construct features from the fMRI data effectively, and improve the classification performance of original features.
Key words:functional magnetic resonance imaging(fMRI); feature engineering; siamese network(SANet) model; siamese network; incremental feature selection
0 引 言
隨著功能磁共振成像(fMRI: functional Magnetic Resonance Imaging)技術(shù)日益成熟, 圍繞fMRI數(shù)據(jù)所展開的研究也逐漸增多。相比其他的腦成像技術(shù), fMRI具有更高的分辨率, 并且通過分析腦區(qū)在一段時(shí)間序列下的血氧飽合度依賴的對(duì)比進(jìn)行研究, 是一種極其高效的研究方法。但fMRI數(shù)據(jù)存在獲取技術(shù)難度大、 對(duì)受試者要求高并且存在頭部運(yùn)動(dòng)與偽影等影響、 數(shù)據(jù)維度高等困難因素。因此, 腦病功能磁共振成像是值得深入探討的課題。為此筆者提出一種基于深度學(xué)習(xí)的fMRI特征構(gòu)造算法, 充分利用孿生網(wǎng)絡(luò)和AlexNet網(wǎng)絡(luò)的結(jié)構(gòu)優(yōu)勢, 并結(jié)合特征工程的長處, 構(gòu)造的fMRI特征在一定程度上提升了分類效果。
1 研究現(xiàn)狀
目前功能性磁共振成像(fMRI)已成為神經(jīng)生理學(xué)和臨床研究中探尋大腦-行為聯(lián)系的最流行的工具之一, 因此以fMRI數(shù)據(jù)為基礎(chǔ)的研究也越來越多。……