Rapid Screening of Carotid Plaque in Cloud Handheld Ultrasound System Based on 5G and AI Technology

doi:10.37015/AUDT.2023.230018

Abstract

Abstract:

Objective: To evaluate the real-time accuracy of cloud handheld ultrasound system using AI technology in screening carotid plaque.

Methods: 2627 ultrasound images of the carotid artery are collected using the cloud handheld system. Bounding boxes of carotid plaques are labeled by qualified sonographers, and the dataset is trained using a lightweight YOLOv3 model. An additional and separate 390 images are collected and tested using the evaluation metrics average recall (AR), average precision (AP), and frames per second (FPS) for quantifying classification performance and time consumption.

Results: We use a plaque grading definition with a thickness of 1.2-1.5 mm defined as small plaque, 1.5-3 mm as medium plaque, and more than 3 mm thick as large plaque. Our model achieves AP^IoU=0.50 with 96.5%, with AP^large is 79.9%, AP^medium is 90.7%, AP^small is 93.5%; AR^IoU=0.50 is 64.5%, where AR^large is 60.6%, AR^medium is 58.3%, AR^small is 70.8%, and FPS is 33.3.

Conclusion: We establish a framework for data set construction, model selection, training, and testing of carotid ultrasound images and verify the effectiveness of real-time AI technology in the automatic detection of carotid artery plaque.

Key words: Handheld ultrasound; Carotid plaque; YOLOv3; Artificial intelligence (AI)

Wenjun Zhang, MD, Mi Zhou, PhD, Qingguo Meng, MD, Lin Zhang, MS, Xin Liu, MS, Paul Liu, PhD, Dong Liu, PhD. Rapid Screening of Carotid Plaque in Cloud Handheld Ultrasound System Based on 5G and AI Technology. Advanced Ultrasound in Diagnosis and Therapy, 2023, 7(2): 152-157.

Figures/Tables 7

Table 1

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Table 2

References 33

[1]	Cuadrado-Godia E, Srivastava SK, Saba L, Araki T, Suri HS, Giannopolulos A, et al. Geometric total plaque area is an equally powerful phenotype compared with carotid intima-media thickness for stroke risk assessment: a deep learning approach. Journal for Vascular Ultrasound 2018; 42:162-188. doi: 10.1177/1544316718806421
[2]	Ning B, Zhang D, He W, Wang LS, Jin ZQ. A study on distribution features of neovascularization in atherosclerotic carotid artery plaques: comparing contrast-enhanced ultrasound with histopathology. Ultrasonic Imaging 2019; 41:115-125. doi: 10.1177/0161734618824941
[3]	Vancraeynest D, Pasquet A, Roelants V, Gerber BL, Vanoverschelde JL. Imaging the vulnerable plaque. J Am Coll Cardiol 2011; 57:1961-1979. doi: 10.1016/j.jacc.2011.02.018 pmid: 21565634
[4]	Xiao X, Xue L, Ye L, Li H, He Y. Health care cost and benefits of artificial intelligence-assisted population-based glaucoma screening for the elderly in remote areas of China: a cost-offset analysis. BMC Public Health 2021; 21:1065. doi: 10.1186/s12889-021-11097-w pmid: 34088286
[5]	Khanna NN, Maindarkar MA, Viswanathan V, Fernandes JFE, Paul S, Bhagawati M, et al. Economics of artificial intelligence in healthcare: diagnosis vs. treatment. Healthcare (Basel) 2022; 10:2493.
[6]	Jamthikar AD, Gupta D, Saba L, Khanna NN, Viskovic K, Mavrogeni S, et al. Artificial intelligence framework for predictive cardiovascular and stroke risk assessment models: a narrative review of integrated approaches using carotid ultrasound. Comput Biol Med 2020; 126:104043. doi: 10.1016/j.compbiomed.2020.104043
[7]	Jain PK, Sharma N, Saba L, Paraskevas KI, Kalra MK, Johri A, et al. Unseen artificial intelligence-deep learning paradigm for segmentation of low atherosclerotic plaque in carotid ultrasound: a multicenter cardiovascular study. Diagnostics (Basel) 2021; 11:2257.
[8]	Biswas M, Saba L, Chakrabartty S, Khanna NN, Song H, Suri HS, et al. Two-stage artificial intelligence model for jointly measurement of atherosclerotic wall thickness and plaque burden in carotid ultrasound: a screening tool for cardiovascular/stroke risk assessment. Comput Biol Med 2020; 123:103847. doi: 10.1016/j.compbiomed.2020.103847
[9]	Biswas M, Saba L, Omerzu T, Johri AM, Khanna NN, Viskovic K, et al. A review on joint carotid intima-media thickness and plaque area measurement in ultrasound for cardiovascular/stroke risk monitoring: artificial intelligence framework. J Digit Imaging 2021; 34:581-604. doi: 10.1007/s10278-021-00461-2 pmid: 34080104
[10]	J. Redmon and A. Farhadi, “YOLOv3: an incremental improvement,” arXiv:1804.02767, Apr. 2018
[11]	Wang YJ, Li ZX, Gu HQ, Zhai Y, Jiang Y, Zhao XQ, et al. China Stroke Statistics 2019 Writing Committee. China Stroke Statistics 2019:a report from the National Center for Healthcare Quality Management in Neurological Diseases, China National Clinical Research Center for Neurological Diseases, the Chinese Stroke Association, National Center for Chronic and Non-communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention and Institute for Global Neuroscience and Stroke Collaborations. Stroke Vasc Neurol 2020; 5:211-239.
[12]	el-Barghouty N, Geroulakos G, Nicolaides A, Androulakis A, Bahal V. Computer-assisted carotid plaque characterisation. Eur J Vasc Endovasc Surg 1995; 9:389-393. doi: 10.1016/S1078-5884(05)80005-X
[13]	Lal BK, Hobson RW 2nd, Pappas PJ, Kubicka R, Hameed M, Chakhtoura EY, et al. Pixel distribution analysis of B-mode ultrasound scan images predicts histologic features of atherosclerotic carotid plaques. J Vasc Surg 2002; 35:1210-1217. doi: 10.1067/mva.2002.122888 pmid: 12042733
[14]	Mougiakakou SG, Golemati S, Gousias I, Nicolaides AN, Nikita KS. Computer-aided diagnosis of carotid atherosclerosis based on ultrasound image statistics, laws' texture and neural networks. Ultrasound Med Biol 2007; 33:26-36. pmid: 17189044
[15]	Pazinato DV, Stein BV, de Almeida WR, Werneck Rde O, Mendes Júnior PR, Penatti OA, et al. Pixel-level tissue classification for ultrasound images. IEEE J Biomed Health Inform 2016; 20:256-67. doi: 10.1109/JBHI.2014.2386796
[16]	Lekadir K, Galimzianova A, Betriu A, Del Mar Vila M, Igual L, Rubin DL, et al. A convolutional neural network for automatic characterization of plaque composition in carotid ultrasound. IEEE J Biomed Health Inform 2017; 21:48-55. doi: 10.1109/JBHI.2016.2631401
[17]	Geroulakos G, Sabetai M. Ultrasonic carotid plaque morphology. Arch Hell Med 2000; 17:141-145.
[18]	Rajendra Acharya U, Rama Krishnan MM, Vinitha Sree S, Sanches J, Shafique S, Nicolaides A, et al. Plaque tissue characterization and classification in ultrasound carotid scans: a paradigm for vascular feature amalgamation. IEEE Transactions on Instrumentation & Measurement 2013; 62:392-400.
[19]	Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Communication of the ACM 2017; 60:84-90 doi: 10.1145/3065386
[20]	Dong Y, Pan Y, Zhao X, Li R, Yuan C, Xu W. Identifying carotid plaque composition in mri with convolutional neural networks. IEEE International Conference on Smart Computing (SMARTCOMP) 2017; 2017:1-8.
[21]	He KM, Zhang XY, Ren SQ, Sun J. Deep residual learning for image recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2016; 2016:770-778.
[22]	Ma W, Zhou R, Zhao Y, Xia YJ, Fenster A, Ding MY, et al. Plaque recognition of carotid ultrasound images based on deep residual network. 2019 IEEE 8th Joint International Information Technology and Artificial Intelligence Conference (ITAIC). IEEE, 2022; 41:9-23.
[23]	Jain PK, Sharma N, Saba L, Paraskevas KI, Kalra MK, Johri A, et al. Automated deep learning-based paradigm for high-risk plaque detection in B-mode common carotid ultrasound scans: an asymptomatic Japanese cohort study. Int Angiol 2022; 41:9-23.
[24]	Lin Y, Huang J, Chen Y, Chen Q, Li Z, Cao Q. Intelligent segmentation of intima-media and plaque recognition in carotid artery ultrasound images. Ultrasound Med Biol 2022; 48:469-479. doi: 10.1016/j.ultrasmedbio.2021.11.001
[25]	Vila MDM, Remeseiro B, Grau M, Elosua R, Betriu À, Fernandez-Giraldez E, et al. Semantic segmentation with DenseNets for carotid artery ultrasound plaque segmentation and CIMT estimation. Artif Intell Med 2020; 103:101784. doi: 10.1016/j.artmed.2019.101784
[26]	Ronneberger O, Fischer P, Brox T. U-Net: U-Net: convolutional networks for biomedical image segmentation arXiv preprint arXiv:1505.04597.
[27]	Meshram NH, Mitchell CC, Wilbrand S, Dempsey RJ, Varghese T. Deep learning for carotid plaque segmentation using a dilated U-Net architecture. Ultrason Imaging 2020; 42:221-230. doi: 10.1177/0161734620951216 pmid: 32885739
[28]	Zhou R, Azarpazhooh MR, Spence JD, Hashemi S, Ma W, Cheng X, et al. Deep learning-based carotid plaque segmentation from B-mode ultrasound images. Ultrasound Med Biol 2021; 47:2723-2733. doi: 10.1016/j.ultrasmedbio.2021.05.023 pmid: 34217560
[29]	Deng C, Adu J, Xie S, Li Z, Meng Q, Zhang Q, et al. Automatic segmentation of ultrasound images of carotid atherosclerotic plaque based on Dense-UNet. Technol Health Care 2023; 31:165-179. doi: 10.3233/THC-220152
[30]	Arsenescu T, Chifor R, Marita T, Santoma A, Lebovici A, Duma D, et al. 3D ultrasound reconstructions of the carotid artery and thyroid gland using artificial-intelligence-based automatic segmentation-qualitative and quantitative evaluation of the segmentation results via comparison with CT angiography. Sensors (Basel) 2023; 23:2806. doi: 10.3390/s23052806
[31]	Jain PK, Dubey A, Saba L, Khanna NN, Laird JR, Nicolaides A, et al. Attention-based UNet deep learning model for plaque segmentation in carotid ultrasound for stroke risk stratification: an artificial intelligence paradigm. J Cardiovasc Dev Dis 2022; 9:326.
[32]	Zhou R, Guo F, Azarpazhooh MR, Hashemi S, Cheng X, Spence JD, et al. Deep learning-based measurement of total plaque area in B-mode ultrasound images. IEEE J Biomed Health Inform 2021; 25:2967-2977. doi: 10.1109/JBHI.2021.3060163
[33]	Gago L, Vila MDM, Grau M, Remeseiro B, Igual L. An end-to-end framework for intima media measurement and atherosclerotic plaque detection in the carotid artery. Comput Methods Programs Biomed 2022; 223:106954. doi: 10.1016/j.cmpb.2022.106954

Years	0-40	40-50	50-60	60-70	70-80	80-90
<1.5 mm	1.95	2.6	9.51	8.33	7.81	2.86
<2 mm	0.26	0.23	3.52	2.34	2.99	2.34
<3 mm	0	0.13	0.52	0.78	1.17	0.78

	Small	Medium	Large	All
AP^{IoU= 0.5:0.95}	68.6	67.0	53.8	78.6
AR^{IoU= 0.5:0.95}	29.4	27.2	29.2	28.5
AP^{IoU= 0.5}	93.5	90.7	79.9	96.5
AP^{IoU= 0.5}	70.8	58.3	60.6	64.5
AP^{IoU= 0.75}	77.2	66.6	61.5	87.4
AP^{IoU= 0.75}	16.5	11.9	24.2	15.9