進階搜尋


   電子論文尚未授權公開,紙本請查館藏目錄
(※如查詢不到或館藏狀況顯示「閉架不公開」,表示該本論文不在書庫,無法取用。)
系統識別號 U0026-1208202018041300
論文名稱(中文) 透過深度學習自動量化心內膜脂肪以輔助分析心血管疾病之風險
論文名稱(英文) Deep Learning-based Epicardial Adipose Tissue Quantification System for Cardiovascular Disease Risk Analysis
校院名稱 成功大學
系所名稱(中) 資訊工程學系
系所名稱(英) Institute of Computer Science and Information Engineering
學年度 108
學期 2
出版年 109
研究生(中文) 吳昭儀
研究生(英文) Chao-Yi Wu
學號 P76074672
學位類別 碩士
語文別 英文
論文頁數 52頁
口試委員 指導教授-蔣榮先
口試委員-蔡依珊
口試委員-郝沛毅
口試委員-林鵬展
口試委員-李宗儒
中文關鍵字 心內膜脂肪  影像分割  心血管疾病  機器學習  預防醫學  醫療影像 
英文關鍵字 Epicardial Adipose Tissue  image segmentation  machine learning  deep learning  risk analysis  preventive health 
學科別分類
中文摘要 心血管疾病是全球最常見的死亡原因之一,其危險因子如是否罹患糖尿病、抽菸、肥胖等,常被使用於評估罹患風險,因此頗受預防性醫學所重視。近年來逐漸有研究發現心內膜脂肪與心血管疾病有密切關係,主要透過活體組織切片實際分析該脂肪,少部分如鈣化指數在醫療影像中進行觀察可得,可用於分析血管狹窄程度。鑒於近年來深度學習在醫療影像的成功,本研究將以CT影像為輸入資料,採取深度學習模型自動擷取並量化心內膜脂肪,並進一步探討與心血管疾病之關聯性,驗證心內膜脂肪組織是否可成為危險因子之一,以輔助醫師進行心血管疾病的風險分析。
我們基於U-Net架構建立模型從CT影像中擷取心內膜脂肪,其中採取ResNeXt模組進行特徵捲積,並在skip connection上加入scSE機制,將encoder的資訊以兩種注意力機制加強後融合一起,再以後處理方式將心內膜脂肪量化成體積,同時為減少個體之間身形差異的因素,使用BSA對心內膜脂肪體積進行校正。在實驗中,本研究收集了成大醫院4954位病人之CT影像資料與11種臨床病史,在分割心內膜脂肪任務上,我們分別嘗試了弱監督式學習與監督式學習等10種以上的模型,最終選用基於ResNeXt加上scSE 機制的模型,並在DSC達到0.89。
經由實驗結果我們實現心內膜脂肪體積與11種疾病的分析,在p值觀察結果共發現在常見4種慢性病與6種心血管疾病上,心內膜脂肪體積有顯著差異。為進一步比較心內膜脂肪體積與鈣化指數對於冠狀動脈疾病之影響,將鈣化指數大於0的病人過濾,並在考慮共病情況下計算勝算比與p值,發現心內膜脂肪體積對於冠狀動脈疾病有顯著差別影響,並以勝算比1.004顯示越多的心內膜脂肪可能導致越高的機率罹患該疾病,可望成為臨床醫師在診斷時對於病患罹患冠狀動脈疾病一項指標。
英文摘要 Cardiovascular disease is the most common cause of death worldwide, always ranked in the top 3 of the leading causes of death in Taiwan. The importance of risk factors for cardiovascular disease has been attached great significance. Recently, studies focused on fat around arterial vessels inside the pericardium, the epicardial adipose tissue. In our research, we associate epicardial adipose tissue with cardiovascular diseases to find evidence that may support epicardial adipose tissue as a new risk factor.
To obtain epicardial adipose tissue rapidly, we design a system aim to automatically extract the fat tissue and provide the quantified information to doctors. Inspired by the success of deep learning model on image tasks, we build our segmentation model based on U-net and did some data augmentation. With scSE mechanism applied on skip connection and SE-RestNeXt block as the backbone. Our system achieved over 0.89 Dice similarity coefficient on the test data. Depends on the data augmentation, we can perform well on enlarged CT scan images and non-contrast CT scan images.
We conduct our experiment in contrasted CT scan image data from NCKUH 4954 patients. There are 11 types of diseases record and several basic personal information included in the data. To eliminate the inherent difference of each shape, we firstly revise the fat volume with the body surface area. We analyze the relation between epicardial adipose tissue volume and diseases with p-value. We found that diabetes, hyperlipidemia, hypertension, chronic kidney disease, stroke, arrhythmia, acute myocardial infarction, heart failure, peripheral artery occlusive disease, and coronary artery disease are significantly different. As for coronary artery disease with zero calcium score, the odds ratio of EAT volume shows 1.004 with p-value<0.5, leading more EAT volume may cause higher risk. Thus, providing EAT volume can improve preventive health and help doctors diagnosis when the calcium score equals 0.
論文目次 中文摘要 I
Abstract III
誌謝 V
Contents VI
List of Figures VIII
List of Tables XI
Chapter 1 Introduction 1
1.1 Background 1
1.2 Motivation 3
1.3 Research Objectives 4
1.4 Thesis Organization 5
Chapter 2 Related Work 6
2.1 Risk Factors of Cardiovascular Disease 6
2.2 Epicardial Adipose Tissue in CT scan 7
2.3 Deep Learning in Medical Image 8
Chapter 3 Epicardial Adipose Tissue Segmentation 10
3.1 Preliminary Study 10
3.1.1 Weakly-supervised Learning 11
3.1.2 Supervised Learning 13
3.2 EAT Segmentation Model 17
3.2.1 One-stage Strategy 17
3.2.2 Two-stage Strategy 18
3.3 2D pixel to 3D volume 21
3.4 Body Surface Area Revision 22
Chapter 4 Experiments 23
4.1 Experimental Design 23
4.2 Data Description 24
4.3 Evaluation 26
4.4 EAT Extraction 27
4.4.1 Data Preprocess 27
4.4.2 EAT Segmentation 29
4.5 Case Study 32
4.6 Different Thickness and Scale Effect 35
4.7 Relation between BSA modified EAT Volume and Disease 39
Chapter 5 Conclusions and Future Works 46
5.1 Conclusions 46
5.2 Future Works 47
Reference 48
參考文獻 108年國人死因統計結果, 2020.
Anthimopoulos, M., Christodoulidis, S., Ebner, L., Christe, A., &Mougiakakou, S. "Lung pattern classification for interstitial lung diseases using a deep convolutional neural network". IEEE Transactions on Medical Imaging, 35(5), 1207–1216, 2016.
Avendi, M. R., Kheradvar, A., &Jafarkhani, H. "A combined deep-learning and deformable-model approach to fully automatic segmentation of the left ventricle in cardiac MRI". Medical Image Analysis, 30, 108–119, 2016.
Barbosa, J. G., Figueiredo, B., Bettencourt, N., &Tavares, J. M. R. S. "Towards automatic quantification of the epicardial fat in non-contrasted CT images". Computer Methods in Biomechanics and Biomedical Engineering, 14(10), 905–914, 2011.
Bertaso, A. G., Bertol, D., Duncan, B. B., &Foppa, M. "Epicardial fat: Definition, measurements and systematic review of main outcomes". Arquivos Brasileiros de Cardiologia, 101(1), 18–28, 2013.
Bolya, D., Zhou, C., Xiao, F., &Lee, Y. J. "Yolact: Real-time instance segmentation". Proceedings of the IEEE International Conference on Computer Vision, 9157–9166, 2019.
Chollet, F. "Xception: Deep learning with depthwise separable convolutions". Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 1251–1258, 2017.
Commandeur, F., Goeller, M., Betancur, J., Cadet, S., Doris, M., Chen, X., …Dey, D. "Deep learning for quantification of epicardial and thoracic adipose tissue from non-contrast CT". IEEE Transactions on Medical Imaging, 37(8), 1835–1846, 2018.
D’agostino, R. B., Vasan, R. S., Pencina, M. J., Wolf, P. A., Cobain, M., Massaro, J. M., &Kannel, W. B. "General cardiovascular risk profile for use in primary care". Circulation, 117(6), 743–753, 2008.
Ding, X., Terzopoulos, D., Diaz-Zamudio, M., Berman, D. S., Slomka, P. J., &Dey, D. "Automated epicardial fat volume quantification from non-contrast CT". Medical Imaging 2014: Image Processing, 9034, 90340I, 2014.
Ding, X., Terzopoulos, D., Diaz-Zamudio, M., Berman, D. S., Slomka, P. J., &Dey, D. "Automated pericardium delineation and epicardial fat volume quantification from noncontrast CT". Medical Physics, 42(9), 5015–5026, 2015.
Goeller, M., Achenbach, S., Marwan, M., Doris, M. K., Cadet, S., Commandeur, F., …others. "Epicardial adipose tissue density and volume are related to subclinical atherosclerosis, inflammation and major adverse cardiac events in asymptomatic subjects". Journal of Cardiovascular Computed Tomography, 12(1), 67–73, 2018.
Havaei, M., Davy, A., Warde-Farley, D., Biard, A., Courville, A., Bengio, Y., …Larochelle, H. "Brain tumor segmentation with deep neural networks". Medical Image Analysis, 35, 18–31, 2017.
He, K., Zhang, X., Ren, S., &Sun, J. "Deep residual learning for image recognition". Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2016-Decem, 770–778, 2016.
Hirata, Y., Kurobe, H., Akaike, M., Chikugo, F., Hori, T., Bando, Y., …Sata, M. "Enhanced inflammation in epicardial fat in patients with coronary artery disease". International Heart Journal, 52(3), 139–142, 2011.
Hu, J., Shen, L., &Sun, G. "Squeeze-and-excitation networks". Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 7132–7141, 2018.
Iacobellis, G., Corradi, D., &Sharma, A. M. "Epicardial adipose tissue: anatomic, biomolecular and clinical relationships with the heart". Nature Clinical Practice Cardiovascular Medicine, 2(10), 536–543, 2005.
Irvin, J., Rajpurkar, P., Ko, M., Yu, Y., Ciurea-Ilcus, S., Chute, C., …others. "Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison". Proceedings of the AAAI Conference on Artificial Intelligence, 33, 590–597, 2019.
Krizhevsky, A., Sutskever, I., &Hinton, G. E. "Imagenet classification with deep convolutional neural networks". Advances in Neural Information Processing Systems, 1097–1105, 2012.
Li, X., Chen, H., Qi, X., Dou, Q., Fu, C. W., &Heng, P. A. "H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes". IEEE Transactions on Medical Imaging, 37(12), 2663–2674, 2018.
Lin, T. Y., Dollár, P., Girshick, R., He, K., Hariharan, B., &Belongie, S. "Feature pyramid networks for object detection". Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, 2017-Janua, 936–944, 2017.
Mahabadi, A. A., Berg, M. H., Lehmann, N., Kälsch, H., Bauer, M., Kara, K., …others. "Association of epicardial fat with cardiovascular risk factors and incident myocardial infarction in the general population: the Heinz Nixdorf Recall Study". Journal of the American College of Cardiology, 61(13), 1388–1395, 2013.
Nakazato, R., Shmilovich, H., Tamarappoo, B. K., Cheng, V. Y., Slomka, P. J., Berman, D. S., &Dey, D. "Interscan reproducibility of computer-aided epicardial and thoracic fat measurement from noncontrast cardiac CT". Journal of Cardiovascular Computed Tomography, 5(3), 172–179, 2011.
Pereira, S., Pinto, A., Alves, V., &Silva, C. A. "Brain tumor segmentation using convolutional neural networks in MRI images". IEEE Transactions on Medical Imaging, 35(5), 1240–1251, 2016.
Pletcher, M. J., Tice, J. A., Pignone, M., &Browner, W. S. "Using the coronary artery calcium score to predict coronary heart disease events: a systematic review and meta-analysis". Archives of Internal Medicine, 164(12), 1285–1292, 2004.
Redmon, J., &Farhadi, A. "YOLO9000: Better, faster, stronger". Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, 2017-Janua, 6517–6525, 2017.
Ronneberger, O., Fischer, P., &Brox, T. "U-net: Convolutional networks for biomedical image segmentation". International Conference on Medical Image Computing and Computer-Assisted Intervention, 234–241, 2015.
Roy, A. G., Navab, N., &Wachinger, C. "Concurrent spatial and channel ‘squeeze & excitation’in fully convolutional networks". International Conference on Medical Image Computing and Computer-Assisted Intervention, 421–429, 2018.
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., …others. "Imagenet large scale visual recognition challenge". International Journal of Computer Vision, 115(3), 211–252, 2015.
Shelhamer, E., Long, J., &Darrell, T. "Fully Convolutional Networks for Semantic Segmentation". IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(4), 640–651, 2017.
Shen, W., Zhou, M., Yang, F., Yu, D., Dong, D., Yang, C., …Tian, J. "Multi-crop convolutional neural networks for lung nodule malignancy suspiciousness classification". Pattern Recognition, 61, 663–673, 2017.
Szegedy, C., Ioffe, S., Vanhoucke, V., &Alemi, A. A. "Inception-v4, inception-resnet and the impact of residual connections on learning". Thirty-First AAAI Conference on Artificial Intelligence, 2017.
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., &Wojna, Z. "Rethinking the inception architecture for computer vision". Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2818–2826, 2016.
Xie, S., Girshick, R., Dollár, P., Tu, Z., &He, K. "Aggregated residual transformations for deep neural networks". Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 1492–1500, 2017.
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2022-08-08起公開。
  • 同意授權校外瀏覽/列印電子全文服務,於2022-08-08起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw