Automatic detection of A-line in lung ultrasound images using deep learning and image processing.
A-line detection
Faster R-CNN
TV filter
gray difference
matched filter
Journal
Medical physics
ISSN: 2473-4209
Titre abrégé: Med Phys
Pays: United States
ID NLM: 0425746
Informations de publication
Date de publication:
Jan 2023
Jan 2023
Historique:
revised:
29
06
2022
received:
24
02
2022
accepted:
30
07
2022
pubmed:
12
8
2022
medline:
25
1
2023
entrez:
11
8
2022
Statut:
ppublish
Résumé
Auxiliary diagnosis and monitoring of lung diseases based on lung ultrasound (LUS) images is important clinical research. A-line is one of the most common indicators of LUS that can offer support for the assessment of lung diseases. A traditional A-line detection method mainly relies on experienced clinicians, which is inefficient and cannot meet the needs of these areas with backward medical level. Therefore, how to realize the automatic detection of A-line in LUS image is important. In order to solve the disadvantages of traditional A-line detection methods, realize automatic and accurate detection, and provide theoretical support for clinical application, we proposed a novel A-line detection method for LUS images with different probe types in this paper. First, the improved Faster R-CNN model with a selection strategy of localization box was designed to accurately locate the pleural line. Then, the LUS image below the pleural line was segmented for independent analysis excluding the influence of other similar structures. Next, image-processing methods based on total variation, matched filter, and gray difference were applied to achieve the automatic A-line detection. Finally, the "depth" index was designed to verify the accuracy by judging whether the automatic measurement results belong to corresponding manual results (±5%). In experiments, 3000 convex array LUS images were used for training and validating the improved pleural line localization model by five-fold cross validation. 850 convex array LUS images and 1080 linear array LUS images were used for testing the trained pleural line localization model and the proposed image-processing-based A-line detection method. The accuracy analysis, error statistics, and Harsdorff distance were employed to evaluate the experimental results. After 100 epochs, the mean loss value of training and validation set of improved Faster R-CNN model reached 0.6540 and 0.7882, with the validation accuracy of 98.70%. The trained pleural line localization model was applied in the testing set of convex and linear probes and reached the accuracy of 97.88% and 97.11%, respectively, which were 3.83% and 8.70% higher than the original Faster R-CNN model. The accuracy, sensitivity, and specificity of A-line detection reached 95.41%, 0.9244%, 0.9875%, and 94.63%, 0.9230%, and 0.9766% for convex and linear probes, respectively. Compared to the experienced clinicians' results, the mean value and p value of depth error were 1.5342 ± 1.2097 and 0.9021, respectively, and the Harsdorff distance was 5.7305 ± 1.8311. In addition, the accumulated accuracy of the two-stage experiment (pleural line localization and A-line detection) was calculated as the final accuracy of the whole A-line detection system. They were 93.39% and 91.90% for convex and linear probes, respectively, which were higher than these previous methods. The proposed method combining image processing and deep learning can automatically and accurately detect A-line in LUS images with different probe types, which has important application value for clinical diagnosis.
Sections du résumé
BACKGROUND
BACKGROUND
Auxiliary diagnosis and monitoring of lung diseases based on lung ultrasound (LUS) images is important clinical research. A-line is one of the most common indicators of LUS that can offer support for the assessment of lung diseases. A traditional A-line detection method mainly relies on experienced clinicians, which is inefficient and cannot meet the needs of these areas with backward medical level. Therefore, how to realize the automatic detection of A-line in LUS image is important.
PURPOSE
OBJECTIVE
In order to solve the disadvantages of traditional A-line detection methods, realize automatic and accurate detection, and provide theoretical support for clinical application, we proposed a novel A-line detection method for LUS images with different probe types in this paper.
METHODS
METHODS
First, the improved Faster R-CNN model with a selection strategy of localization box was designed to accurately locate the pleural line. Then, the LUS image below the pleural line was segmented for independent analysis excluding the influence of other similar structures. Next, image-processing methods based on total variation, matched filter, and gray difference were applied to achieve the automatic A-line detection. Finally, the "depth" index was designed to verify the accuracy by judging whether the automatic measurement results belong to corresponding manual results (±5%). In experiments, 3000 convex array LUS images were used for training and validating the improved pleural line localization model by five-fold cross validation. 850 convex array LUS images and 1080 linear array LUS images were used for testing the trained pleural line localization model and the proposed image-processing-based A-line detection method. The accuracy analysis, error statistics, and Harsdorff distance were employed to evaluate the experimental results.
RESULTS
RESULTS
After 100 epochs, the mean loss value of training and validation set of improved Faster R-CNN model reached 0.6540 and 0.7882, with the validation accuracy of 98.70%. The trained pleural line localization model was applied in the testing set of convex and linear probes and reached the accuracy of 97.88% and 97.11%, respectively, which were 3.83% and 8.70% higher than the original Faster R-CNN model. The accuracy, sensitivity, and specificity of A-line detection reached 95.41%, 0.9244%, 0.9875%, and 94.63%, 0.9230%, and 0.9766% for convex and linear probes, respectively. Compared to the experienced clinicians' results, the mean value and p value of depth error were 1.5342 ± 1.2097 and 0.9021, respectively, and the Harsdorff distance was 5.7305 ± 1.8311. In addition, the accumulated accuracy of the two-stage experiment (pleural line localization and A-line detection) was calculated as the final accuracy of the whole A-line detection system. They were 93.39% and 91.90% for convex and linear probes, respectively, which were higher than these previous methods.
CONCLUSIONS
CONCLUSIONS
The proposed method combining image processing and deep learning can automatically and accurately detect A-line in LUS images with different probe types, which has important application value for clinical diagnosis.
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
330-343Subventions
Organisme : National Natural Science Foundation of China
ID : 12034005
Organisme : National Natural Science Foundation of China
ID : 61975056
Organisme : Natural Science Foundation of Shanghai
ID : 19ZR1416000
Organisme : Science and Technology Commission of Shanghai Municipality
ID : 20440713100
Organisme : Science and Technology Commission of Shanghai Municipality
ID : 18DZ2270800
Organisme : Shanghai Municipal Health Commission
ID : GWV-10.1-XK24
Organisme : Shanghai Shenkang Hospital Management
ID : SHDC2020RC6030-002
Organisme : CAMS Innovation Fund for Medical Sciences
ID : 2021-I2M-C&T-B-015
Organisme : Shanghai Science and Technology Plan Project
ID : 21Y11902500
Organisme : Shanghai Science and Technology Commission
ID : 20Z11900904
Informations de copyright
© 2022 American Association of Physicists in Medicine.
Références
Chen J, He C, Yin J, et al. Quantitative analysis and automated lung ultrasound scoring for evaluating COVID-19 pneumonia with neural networks. IEEE Trans Ultrason Ferroelectr Freq Control. 2021;68(7):2507-2515.
Mayo PH, Copetti R, Feller-Kopman D, et al. Thoracic ultrasonography: a narrative review. Intensive Care Med. 2019;45(9):1200-1211.
Bouhemad B, Brisson H, Le-Guen M, Arbelot C, Lu Q, Rouby JJ. Bedside ultrasound assessment of positive end-expiratory pressure-induced lung recruitment. Am J Respir Crit Care Med. 2011;183(3):341-347.
Soldati G, Smargiassi A, Inchingolo R, et al. Proposal for international standardization of the use of lung ultrasound for patients with COVID-19: a simple, quantitative, reproducible method. J Ultrasound Med. 2020;39(7):1413-1419.
Allinovi M, Parise A, Giacalone M, et al. Lung ultrasound may support diagnosis and monitoring of COVID-19 pneumonia. Ultrasound Med Biol. 2020;46(11):2908-2917.
La Salvia M, Secco G, Torti E, et al. Deep learning and lung ultrasound for Covid-19 pneumonia detection and severity classification. Comput Biol Med. 2021;136:104742-104742.
Pecho-Silva S, Claudia Navarro-Solsol A, Taype-Rondan A, et al. Pulmonary ultrasound in the diagnosis and monitoring of coronavirus disease (COVID-19): a systematic review. Ultrasound Med Biol. 2021;47(8):1997-2005.
Buonsenso D, Raffaelli F, Tamburrini E, et al. Clinical role of lung ultrasound for diagnosis and monitoring of COVID-19 pneumonia in pregnant women. Ultrasound Obstet Gynecol. 2020;56(1):106-109.
Hussain A, Via G, Melniker L, et al. Multi-organ point-of-care ultrasound for COVID-19 (PoCUS4COVID): international expert consensus. Crit Care. 2020;24(1):702.
Ji L, Cao C, Gao Y, et al. Prognostic value of bedside lung ultrasound score in patients with COVID-19. Crit Care. 2020;24(1):700.
Li W, Fu ML, Qian C, et al. Quantitative assessment of COVID-19 pneumonia in neonates using lung ultrasound score. Pediatr Pulmonol. 2021;56:1419-1426.
Mento F, Perrone T, Macioce VN, et al. On the impact of different lung ultrasound imaging protocols in the evaluation of patients affected by coronavirus disease 2019 how many acquisitions are needed? J Ultrasound Med. 2021;40(10):2235-2238.
Yu ACH, Demi L, Muller M, Zhou QF. Ultrasound imaging: a silent hero in COVID-19 and lung diagnostics. IEEE Trans Ultrason Ferroelectr Freq Control. 2020;67(11):2194-2196.
Zhao L, Yu K, Zhao Q, et al. Lung ultrasound score in evaluating the severity of coronavirus disease 2019 (COVID-19) pneumonia. Ultrasound Med Biol. 2020;46(11):2938-2944.
Soldati G, Demi M, Smargiassi A, Inchingolo R, Demi L. The role of ultrasound lung artifacts in the diagnosis of respiratory diseases. Expert Rev Respir Med. 2019;13(2):163-172.
Zhu ST, Tao FY, Xu JH, et al. Utility of point-of-care lung ultrasound for clinical classification of COVID-19. Ultrasound Med Biol. 2021;47(2):214-221.
van Sloun RJG, Demi L. Localizing B-lines in lung ultrasonography by weakly supervised deep learning, in-vivo results. IEEE J Biomed Health Inf. 2020;24(4):957-964.
Brusasco C, Santori G, Bruzzo E, et al. Quantitative lung ultrasonography: a putative new algorithm for automatic detection and quantification of B-lines. Crit Care. 2019;23(1):288.
Brattain LJ, Telfer BA, Liteplo AS, Noble VE. Automated B-line scoring on thoracic sonography. J Ultrasound Med. 2013;32(12):2185-2190.
Moshavegh R, Hansen KL, Moller-Sorensen H, Nielsen MB, Jensen JA. Automatic detection of B-lines in in vivo lung ultrasound. IEEE Trans Ultrason Ferroelectr Freq Control. 2019;66(2):309-317.
Anantrasirichai N, Hayes W, Allinovi M, Bull D, Achim A. Line detection as an inverse problem: application to lung ultrasound imaging. IEEE Trans Med Imaging. 2017;36(10):2045-2056.
Anderson KL, Fields JM, Panebianco NL, et al. Inter-rater reliability of quantifying pleural b-lines using multiple counting methods. J Ultrasound Med. 2013;32(1):115-120.
Roy S, Menapace W, Oei S, et al. Deep learning for classification and localization of COVID-19 markers in point-of-care lung ultrasound. IEEE Trans Med Imaging. 2020;39(8):2676-2687.
Chen J, Li J, He C, Li W, Li Q. Automated pleural line detection based on radon transform using ultrasound. Ultrason Imaging. 2021;43(1):19-28.
Chiesa AM, Ciccarese F, Gardelli G, et al. Sonography of the normal lung: comparison between young and elderly subjects. J Clin Ultrasound. 2015;43(4):230-234.
Carrer L, Donini E, Marinelli D, et al. Automatic pleural line extraction and COVID-19 scoring from lung ultrasound data. IEEE Trans Ultrason Ferroelectr Freq Control. 2020;67(11):2207-2217.
Lichtenstein DA, Meziere GA, Lagoueyte JF, Biderman P, Goldstein I, Gepner A. A-lines and B-lines lung ultrasound as a bedside tool for predicting pulmonary artery occlusion pressure in the critically III. Chest. 2009;136(4):1014-1020.
Karakus O, Anantrasirichai N, Aguersif A, Silva S, Basarab A, Achim A. Detection of line artifacts in lung ultrasound images of COVID-19 patients via nonconvex regularization. IEEE Trans Ultrason Ferroelectr Freq Control. 2020;67(11):2218-2229.
Susanti H, Suprijanto. Image processing framework for pleural line (A-line) detection in video lung ultrasonography. 2020 IEEE-EMBS Conference on Biomedical Engineering and Sciences, Electr Network, IEEE; 2021:99-102.
Gare GR, Schoenling A, Philip V, et al. Dense pixel-labeling for reverse-transfer and diagnostic learning on lung ultrasound for Covid-19 and pneumonia detection. 2021 IEEE 18th International Symposium on Biomedical Imaging, France, IEEE; 2021:1406-1410.
Ren S, He K, Girshick R, Sun J. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell. 2017;39(6):1137-1149.
Wan SH, Goudos S. Faster R-CNN for multi-class fruit detection using a robotic vision system. Comput Netw. 2020;168:6.
Rudin LI, Osher S, Fatemi E. Nonlinear total variation based noise removal algorithms. Physica D. 1992;60(1-4):259-268.
Huang Q, Yang F, Liu L, Li X. Automatic segmentation of breast lesions for interaction in ultrasonic computer-aided diagnosis. Inf Sci. 2015;314, 293-310.
Xing WY, Deng N, Xin BJ, Liu YW, Chen Y, Zhang ZY. Identification of extremely similar animal fibers based on matched filter and HOG-SVM. IEEE Access. 2019;7:98603-98617.
Chaudhuri S, Chatterjee S, Katz N, Nelson M, Goldbaum M. Detection of blood-vessels in retinal images using two-dimensional matched-filters. IEEE Trans Med Imaging. 1989;8(3):263-269.
Liu YW, Liu X, Wang SH, Song JL, Zhang JQ. A novel method for accurate extraction of liver capsule and auxiliary diagnosis of liver cirrhosis based on high-frequency ultrasound images. Comput Biol Med. 2020;125:15.
Zhao WX, Lu M, Wang XM, Guo Y. The role of sarcopenia questionnaires in hospitalized patients with chronic heart failure. Aging Clin Exp Res. 2021;33(2):339-344.
Zhao C, Xu Y, He Z, et al. Lung segmentation and automatic detection of COVID-19 using radiomic features from chest CT images. Pattern recognit. 2021;119:108071.
Alhudhaif A, Polat K, Karaman O. Determination of COVID-19 pneumonia based on generalized convolutional neural network model from chest X-ray images. Expert Syst Appl. 2021;180:9.
Pulgar-Sanchez M, Chamorro K, Fors M, et al. Biomarkers of severe COVID-19 pneumonia on admission using data-mining powered by common laboratory blood tests-datasets. Comput Biol Med. 2021;136:104738-104738.