心血管疾病在我國已連續30年位居十大死因前三名。許多相關疾病都透過心血管造影術進行觀察。然而，由於影像解析度不足，肉眼觀察既主觀又缺乏量化，因此急需先進的技術解決方案。目前流行的人工智慧技術鮮少能在遠端裝置上實際應用，供醫療人員使用。

為解決此問題，本研究的主要目的在於開發一種適用於遠端裝置的心血管造影演算法。我們利用傳統影像處理方法與改良版的Frangi演算法，並成功地將此演算法應用到iOS應用程式的開發中，使其能在iPhone、iPad等裝置上進行離線操作，為醫療人員提供實地使用的解決方案。

通過這項技術，我們成功將演算法轉化為一款用戶友好、易於使用且提供實時反饋的心血管造影應用程式。該應用程式無需大量訓練資料，運行速度快，精確度高，因此解決了目前深度學習模型在訓練和推論過程中面臨的許多挑戰。

在測試階段，我們使用500張圖像（包括來自DCA1數據集的公開圖像數據）來驗證該演算法。結果表明，我們的演算法具有不俗的精確度，甚至與當前可用的先進AI模型的性能相近。這凸顯出我們的演算法在離線操作和實時處理方面的優勢。結果顯示，我們的冠狀動脈造影應用不僅達到預期的性能目標，也為醫學影像處理的未來提供了新的方向。

To better identify the vessels, image segmentation techniques are often applied to Coronary Angiography (CAG) which reveals the functions and structures of heart's arteries using X-Ray images. Although deep learning based segmentation methods have shown their superiority in accuracy, they are often too complex to be used in medical edge computing, a way to provides prompt diagnoses with minimum hardware cost. In this study, we investigate the methods for CAG segmentation on mobile edge devices and propose a novel method, called Adaptive Frangi Filter (AFF). Frangi filter is a classical method for vessel segmentation, but suffers from the problems of long processing time for parameter search and the noisy outputs. The adaptive Frangi filter utilizes a lightweight neural network to recognize the vessel patterns to decide the most suitable parameters. It also employs the statistical information of vessels to remove the noises from the segmented results. We have implemented AFF on mobile devices to demonstrate its usability. Experimental results show that AFF can achieve similar segmentation accuracy to the state-of-the-art models, with a much smaller code size, an efficient training process, and a faster inference time on mobile edge devices.

Contents
Abstract (Chinese) I
Abstract II
Contents III
List of Figures V
List of Tables VI
1 Introduction 1
2 Related Works 5
2.1 CAG Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Edge Computing for Radiology . . . . . . . . . . . . . . . . . . . . 8
3 AFF Method and Implementation 10
3.1 Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1.1 Darker border removal . . . . . . . . . . . . . . . . . . . . . 11
3.1.2 Background structure elimination . . . . . . . . . . . . . . . 11
3.1.3 Contrast enhancement and noise reduction . . . . . . . . . . 12
3.2 Adaptive Frangi Filter . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.1 Training data collection . . . . . . . . . . . . . . . . . . . . 13
3.2.2 Model training . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.3 Scale selection . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.3 Statistical Enhancement . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Implementation on Mobile Devices . . . . . . . . . . . . . . . . . . 18
3.4.1 Packaging into an iOS Application . . . . . . . . . . . . . . 20
3.4.2 MobileNetV3 . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4 Experimentation 25
4.1 Experimental Settings . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Experiment Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2.1 Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Scale Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5 Conclusion and Future Work 32
Bibliography 34

[1] Mohamed Abd Elaziz, Abdelghani Dahou, Naser A Alsaleh, Ammar H
Elsheikh, Amal I Saba, and Mahmoud Ahmadein. Boosting COVID-19 Image
Classification Using MobileNetV3 and Aquila Optimizer Algorithm. Entropy,
23(11):1383, 2021.
[2] Shahab Aslani and Haldun Sarnel. A new supervised retinal vessel segmen-
tation method based on robust hybrid features. Biomedical Signal Processing
and Control, 30:1–12, 2016.
[3] Vijay Badrinarayanan, Alex Kendall, and Roberto Cipolla. SegNet: A Deep
Convolutional Encoder-Decoder Architecture for Image Segmentation. IEEE
Transactions on Pattern Analysis and Machine Intelligence, 39(12):2481–
2495, 2017.
[4] Fernando Cervantes-Sanchez, Ivan Cruz-Aceves, Arturo Hernandez-Aguirre,
and Sergio E. Hernandez-Gonzalez, Martha A.and Solorio-Meza. Automatic
Segmentation of Coronary Arteries in X-ray Angiograms using Multiscale
Analysis and Artificial Neural Networks, 2019.
[5] Yu Cheng, Duo Wang, Pan Zhou, and Tao Zhang. A Survey of Model Com-
pression and Acceleration for Deep Neural Networks, 2020.
[6] Alejandro F. Frangi, Wiro J. Niessen, Koen L. Vincken, and Max A. Viergever.
Multiscale Vessel Enhancement Filtering. In William M. Wells, Alan Colch-
34
ester, and Scott Delp, editors, Medical Image Computing and Computer-
Assisted Intervention — MICCAI’98, pages 130–137, Berlin, Heidelberg,
1998. Springer Berlin Heidelberg.
[7] Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the Knowledge in a
Neural Network, 2015.
[8] A. Howard, M. Sandler, B. Chen, W. Wang, L. Chen, M. Tan, G. Chu, V. Va-
sudevan, Y. Zhu, R. Pang, H. Adam, and Q. Le. Searching for MobileNetV3.
In 2019 IEEE/CVF International Conference on Computer Vision (ICCV),
pages 1314–1324, Los Alamitos, CA, USA, nov 2019. IEEE Computer Society.
[9] Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun
Wang, Tobias Weyand, Marco Andreetto, and Hartwig Adam. MobileNets:
Efficient Convolutional Neural Networks for Mobile Vision Applications, 2017.
[10] Ahsan Khawaja, Tariq M. Khan, Khuram Naveed, Syed Saud Naqvi,
Naveed Ur Rehman, and Syed Junaid Nawaz. An Improved Retinal Vessel
Segmentation Framework Using Frangi Filter Coupled With the Probabilistic
Patch Based Denoiser. IEEE Access, 7:164344–164361, 2019.
[11] Qinbin Li, Zeyi Wen, Zhaomin Wu, Sixu Hu, Naibo Wang, Yuan Li, Xu Liu,
and Bingsheng He. A Survey on Federated Learning Systems: Vision, Hype
and Reality for Data Privacy and Protection. IEEE Transactions on Knowl-
edge and Data Engineering, 35(4):3347–3366, apr 2023.
[12] Liang Liming, Sheng Xiaoqi, Lan Zhimin, and QianYanqun. Retinal Blood
Vessel Segmentation Algorithm Based on Multidirectional Filtering, 2022.
Optics Journal Article.
[13] J. Long, E. Shelhamer, and T. Darrell. Fully convolutional networks for
semantic segmentation. In 2015 IEEE Conference on Computer Vision and
35
Pattern Recognition (CVPR), pages 3431–3440, Los Alamitos, CA, USA, jun
2015. IEEE Computer Society.
[14] João Lourenço-Silva, Miguel Nobre Menezes, Tiago Rodrigues, Beatriz Silva,
Fausto J. Pinto, and Arlindo L. Oliveira. Encoder-Decoder Architectures
for Clinically Relevant Coronary Artery Segmentation. In Mukul S. Bansal,
Ion Măndoiu, Marmar Moussa, Murray Patterson, Sanguthevar Rajasekaran,
Pavel Skums, and Alexander Zelikovsky, editors, Computational Advances in
Bio and Medical Sciences, pages 63–78, Cham, 2022. Springer International
Publishing.
[15] Nobuyuki Otsu. A Threshold Selection Method from Gray-Level Histograms.
IEEE Transactions on Systems, Man, and Cybernetics, 9(1):62–66, 1979.
[16] Devin Partida. The Edge of Medical Imaging: How Edge Computing Is
Changing Radiology. https://www.medtechintelligence.com/column/
the-edge-of-medical-imaging-how-edge-computing-is-changing-radiology/,
July 21, 2021.
[17] Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Brad-
bury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein,
Luca Antiga, Alban Desmaison, Andreas Kopf, Edward Yang, Zachary De-
Vito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner,
Lu Fang, Junjie Bai, and Soumith Chintala. Pytorch: An imperative
style, high-performance deep learning library. In H. Wallach, H. Larochelle,
A. Beygelzimer, F. d'Alché-Buc, E. Fox, and R. Garnett, editors, Advances in
Neural Information Processing Systems, volume 32. Curran Associates, Inc.,
2019.
36
[18] Binjie Qin, Haohao Mao, Yiming Liu, Jun Zhao, Yisong Lv, Yu Ying Zhu,
Song Ding, and Xu Chen. Robust PCA Unrolling Network for Super-
Resolution Vessel Extraction in X-Ray Coronary Angiography. IEEE Trans-
actions on Medical Imaging, 41:3087–3098, 2022.
[19] MIT Technology Review. TRANSFORMING HEALTH CARE AT
THE EDGE. https://www.technologyreview.com/2021/06/10/1026038/
transforming-health-care-at-the-edge/, June 2021.
[20] David Rivest-Hénault, Hari Sundar, and Mohamed Cheriet. Nonrigid 2D/3D
Registration of Coronary Artery Models With Live Fluoroscopy for Guidance
of Cardiac Interventions. IEEE Transactions on Medical Imaging, 31(8):1557–
1572, 2012.
[21] Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-Net: Convolutional
Networks for Biomedical Image Segmentation. In Nassir Navab, Joachim
Hornegger, William M. Wells, and Alejandro F. Frangi, editors, Medical Image
Computing and Computer-Assisted Intervention – MICCAI 2015, pages 234–
241, Cham, 2015. Springer International Publishing.
[22] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L. Chen. MobileNetV2:
Inverted Residuals and Linear Bottlenecks. In 2018 IEEE/CVF Conference
on Computer Vision and Pattern Recognition (CVPR), pages 4510–4520, Los
Alamitos, CA, USA, jun 2018. IEEE Computer Society.
[23] Agung W. Setiawan, Tati R. Mengko, Oerip S. Santoso, and Andriyan B.
Suksmono. Color retinal image enhancement using CLAHE. In International
Conference on ICT for Smart Society, pages 1–3, 2013.
[24] Marwa Shams, Mohammed A. M. Salem, Safwat Hamad, and Howida A.
Shedeed. Coronary artery tree segmentation in computed tomography angiog-
37
raphy using Otsu method. In 2017 Eighth International Conference on Intel-
ligent Computing and Information Systems (ICICIS), pages 416–420, 2017.
[25] Xingxiang Tao, Hao Dang, Xiaoguang Zhou, Xiangdong Xu, and Danqun
Xiong. A Lightweight Network for Accurate Coronary Artery Segmentation
Using X-Ray Angiograms. Frontiers in Public Health, 10, 2022.
[26] Minghan Yan, Jian Zhou, Cong Luo, Tingfa Xu, and Xiaoxue Xing. Multiscale
Joint Optimization Strategy for Retinal Vascular Segmentation. Sensors,
22:1258, 2022.
[27] Su Yang, Jihoon Kweon, Jae-Hyung Roh, Jae-Hwan Lee, Heejun Kang, Lae-
Jeong Park, Dong Jun Kim, Hyeonkyeong Yang, Jaehee Hur, Do-Yoon Kang,
Pil Hyung Lee, Jung-Min Ahn, Soo-Jin Kang, Duk-Woo Park, Seung-Whan
Lee, Young-Hak Kim, Cheol Whan Lee, Seong-Wook Park, and Seung-Jung
Park. Deep learning segmentation of major vessels in X-ray coronary angiog-
raphy. Scientific Reports, 9(1):16897, 2019.
[28] Jingyang Zhang, Guotai Wang, Hongzhi Xie, Shuyang Zhang, Zhenghui Shi,
and Lixu Gu. Vesselness-constrained robust PCA for vessel enhancement in
X-ray coronary angiograms. Physics in medicine and biology, 63(15):155019,
2018.
[29] Min-Ling Zhang and Zhi-Hua Zhou. Multilabel Neural Networks with Appli-
cations to Functional Genomics and Text Categorization. IEEE Transactions
on Knowledge and Data Engineering, 18(10):1338–1351, 2006.
[30] Hengshuang Zhao, Jianping Shi, Xiaojuan Qi, Xiaogang Wang, and Jiaya Jia.
Pyramid Scene Parsing Network. In Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition (CVPR), July 2017.