Machine learning-based segmentation of ischemic penumbra by using diffusion tensor metrics in a rat model

被引:7
作者
Kuo, Duen-Pang [1 ,2 ]
Kuo, Po-Chih [3 ]
Chen, Yung-Chieh [1 ]
Kao, Yu-Chieh Jill [4 ]
Lee, Ching-Yen [6 ,7 ]
Chung, Hsiao-Wen [8 ]
Chen, Cheng-Yu [1 ,4 ,5 ,9 ,10 ,11 ]
机构
[1] Taipei Med Univ Hosp, Dept Med Imaging, 250 Wu Hsing St, Taipei 11031, Taiwan
[2] Taoyuan Armed Forces Gen Hosp, Dept Radiol, Taoyuan, Taiwan
[3] MIT, Inst Med Engn & Sci, 77 Massachusetts Ave, Cambridge, MA 02139 USA
[4] Natl Yang Ming Univ, Dept Biomed Imaging & Radiol Sci, 155,Sec 2,Linong St, Taipei 11221, Taiwan
[5] Taipei Med Univ, Coll Med, Sch Med, Dept Radiol, 250 Wu Hsing St, Taipei 11031, Taiwan
[6] Taipei Med Univ Hosp, TMU Ctr Big Data & Artificial Intelligence Med Im, Taipei, Taiwan
[7] Taipei Med Univ Hosp, TMU Res Ctr Artificial Intelligence Med, Taipei, Taiwan
[8] Natl Taiwan Univ, Grad Inst Biomed Elect & Bioinformat, Taipei, Taiwan
[9] Taipei Med Univ Hosp, Radiogen Res Ctr, 250 Wu Hsing St, Taipei 11031, Taiwan
[10] Taipei Med Univ, Ctr Artificial Intelligence Med, 250 Wu Hsing St, Taipei 11031, Taiwan
[11] Natl Def Med Ctr, Dept Radiol, 250 Wu Hsing St, Taipei 11031, Taiwan
关键词
Machine learning; Diffusion tensor imaging; Ischemic penumbra; Infarct core; CEREBRAL-ARTERY OCCLUSION; PERFUSION-WEIGHTED MRI; ACUTE STROKE; LESION SEGMENTATION; HYPERACUTE STROKE; INFARCT CORE; TISSUE FATE; BLOOD-FLOW; MISMATCH; ONSET;
D O I
10.1186/s12929-020-00672-9
中图分类号
Q2 [细胞生物学];
学科分类号
071009 ; 090102 ;
摘要
Background Recent trials have shown promise in intra-arterial thrombectomy after the first 6-24 h of stroke onset. Quick and precise identification of the salvageable tissue is essential for successful stroke management. In this study, we examined the feasibility of machine learning (ML) approaches for differentiating the ischemic penumbra (IP) from the infarct core (IC) by using diffusion tensor imaging (DTI)-derived metrics. Methods Fourteen male rats subjected to permanent middle cerebral artery occlusion (pMCAO) were included in this study. Using a 7 T magnetic resonance imaging, DTI metrics such as fractional anisotropy, pure anisotropy, diffusion magnitude, mean diffusivity (MD), axial diffusivity, and radial diffusivity were derived. The MD and relative cerebral blood flow maps were coregistered to define the IP and IC at 0.5 h after pMCAO. A 2-level classifier was proposed based on DTI-derived metrics to classify stroke hemispheres into the IP, IC, and normal tissue (NT). The classification performance was evaluated using leave-one-out cross validation. Results The IC and non-IC can be accurately segmented by the proposed 2-level classifier with an area under the receiver operating characteristic curve (AUC) between 0.99 and 1.00, and with accuracies between 96.3 and 96.7%. For the training dataset, the non-IC can be further classified into the IP and NT with an AUC between 0.96 and 0.98, and with accuracies between 95.0 and 95.9%. For the testing dataset, the classification accuracy for IC and non-IC was 96.0 +/- 2.3% whereas for IP and NT, it was 80.1 +/- 8.0%. Overall, we achieved the accuracy of 88.1 +/- 6.7% for classifying three tissue subtypes (IP, IC, and NT) in the stroke hemisphere and the estimated lesion volumes were not significantly different from those of the ground truth(p = .56, .94, and .78, respectively). Conclusions Our method achieved comparable results to the conventional approach using perfusion-diffusion mismatch. We suggest that a single DTI sequence along with ML algorithms is capable of dichotomizing ischemic tissue into the IC and IP.
引用
收藏
页数:11
相关论文
共 55 条
[1]   Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach [J].
Aerts, Hugo J. W. L. ;
Velazquez, Emmanuel Rios ;
Leijenaar, Ralph T. H. ;
Parmar, Chintan ;
Grossmann, Patrick ;
Cavalho, Sara ;
Bussink, Johan ;
Monshouwer, Rene ;
Haibe-Kains, Benjamin ;
Rietveld, Derek ;
Hoebers, Frank ;
Rietbergen, Michelle M. ;
Leemans, C. Rene ;
Dekker, Andre ;
Quackenbush, John ;
Gillies, Robert J. ;
Lambin, Philippe .
NATURE COMMUNICATIONS, 2014, 5
[2]   Implementation and outcome of thrombolysis with alteplase 3-4.5 h after an acute stroke: an updated analysis from SITS-ISTR [J].
Ahmed, Niaz ;
Wahlgren, Nils ;
Grond, Martin ;
Hennerici, Michael ;
Lees, Kennedy R. ;
Mikulik, Robert ;
Parsons, Mark ;
Roine, Risto O. ;
Toni, Danilo ;
Ringleb, Peter .
LANCET NEUROLOGY, 2010, 9 (09) :866-874
[3]   Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging [J].
Albers, G. W. ;
Marks, M. P. ;
Kemp, S. ;
Christensen, S. ;
Tsai, J. P. ;
Ortega-Gutierrez, S. ;
McTaggart, R. A. ;
Torbey, M. T. ;
Kim-Tenser, M. ;
Leslie-Mazwi, T. ;
Sarraj, A. ;
Kasner, S. E. ;
Ansari, S. A. ;
Yeatts, S. D. ;
Hamilton, S. ;
Mlynash, M. ;
Heit, J. J. ;
Zaharchuk, G. ;
Kim, S. ;
Carrozzella, J. ;
Palesch, Y. Y. ;
Demchuk, A. M. ;
Bammer, R. ;
Lavori, P. W. ;
Broderick, J. P. ;
Lansberg, M. G. .
NEW ENGLAND JOURNAL OF MEDICINE, 2018, 378 (08) :708-718
[4]   Ischemic Core and Hypoperfusion Volumes Predict Infarct Size in SWIFT PRIME [J].
Albers, Gregory W. ;
Goyal, Mayank ;
Jahan, Reza ;
Bonafe, Alain ;
Diener, Hans-Christoph ;
Levy, Elad I. ;
Pereira, Vitor M. ;
Cognard, Christophe ;
Cohen, David J. ;
Hacke, Werner ;
Jansen, Olav ;
Jovin, Tudor G. ;
Mattle, Heinrich P. ;
Nogueira, Raul G. ;
Siddiqui, Adnan H. ;
Yavagal, Dileep R. ;
Baxter, Blaise W. ;
Devlin, Thomas G. ;
Lopes, Demetrius K. ;
Reddy, Vivek K. ;
de Rochemont, Richard du Mesnil ;
Singer, Oliver C. ;
Bammer, Roland ;
Saver, Jeffrey L. .
ANNALS OF NEUROLOGY, 2016, 79 (01) :76-89
[5]  
[Anonymous], 2006, PATTERN RECOGN, DOI DOI 10.1117/1.2819119
[6]   Protecting the ischaemic penumbra as an adjunct to thrombectomy for acute stroke [J].
Baron, Jean-Claude .
NATURE REVIEWS NEUROLOGY, 2018, 14 (06) :325-337
[7]   Early identification of potentially salvageable tissue with MRI-based predictive algorithms after experimental ischemic stroke [J].
Bouts, Mark J. R. J. y ;
Tiebosch, Ivo A. C. W. ;
van der Toorn, Annette ;
Viergever, Max A. ;
Wu, Ona ;
Dijkhuizen, Rick M. .
JOURNAL OF CEREBRAL BLOOD FLOW AND METABOLISM, 2013, 33 (07) :1075-1082
[8]   Quantification of perfusion using bolus tracking magnetic resonance imaging in stroke - Assumptions, limitations, and potential implications for clinical use [J].
Calamante, F ;
Gadian, DG ;
Connelly, A .
STROKE, 2002, 33 (04) :1146-1151
[9]   Efficient approximate leave-one-out cross-validation for kernel logistic regression [J].
Cawley, Gavin C. ;
Talbot, Nicola L. C. .
MACHINE LEARNING, 2008, 71 (2-3) :243-264
[10]   Fully automatic acute ischemic lesion segmentation in DWI using convolutional neural networks [J].
Chen, Liang ;
Bentley, Paul ;
Rueckert, Daniel .
NEUROIMAGE-CLINICAL, 2017, 15 :633-643