MRI诊断140例腰椎智能网络自动检测分型MCs方法的比较

doi:10.6040/j.issn.1671-7554.0.2022.1238

摘要/Abstract

摘要： 目的探究使用Yolov5网络检测分类Modic改变(MCs)的性能,与基于Yolov5和Resnet34网络自动检测分类MCs方法进行比较。方法回顾性分析2020年6月至2021年6月接受MRI诊断且符合纳入和排除标准的MCs患者140例,其中男55例,女85例,25~89岁,平均(59.0±13.7)岁。在完成MRI影像的标注工作后,将标注后的MRI影像导入深度学习模型训练,使用医学数据常规增强和Mosaic数据增强进行数据扩充,降低训练数据集过少的因素;利用迁移学习的方法,解决网络在小数据集上过拟合的问题。采用平均精度(AP)、平均精度均值(mAP)、召回率、精确率、F1值等性能指标对两种方法诊断MCs进行评估并比较。结果 Yolov5网络检测分类MCs时,mAP、召回率、精确率和F1值分别达到87.56%、82.05%、89.44%和0.845;Yolov5和Resnet34网络自动检测分类MCs时,召回率、精确率和F1值分别达到88.41%、88.68%和0.885。结论 Yolov5网络可以帮助诊断腰椎MCs,使用Yolov5和Resnet34网络检测分类MCs时,模型诊断MCs的性能提升,进而表明Yolo系列网络可以为智能辅助诊断技术在脊柱领域的应用提供可能性。

关键词: Modic改变, 磁共振成像, 终板炎, Yolov5, Resnet34

Abstract: Objective To explore the performance of Yolov5 network to detect and classify Modic changes(MCs)and to comparethe performance of Yolov5 alone and Yolov5 plus Resnet34. Methods Clinical data of 140 patients with MCs who under went MRI diagnosis and met the inclusion and exclusion criteria during June 2020 and June 2021 were retrospectively analyzed, including 55 males and 85 females, mean age(59.0±13.7)years(range 25-89 years). After the MRI images were labeled, they were imported into the deep learning model, and were expanded by using the conventional enhancement of medical data and Mosaic data enhancement to reduce the effects caused by two few factors of the training data sets. Transfer learning method was used to solve the problem of network overfitting on small data sets. Average precision(AP), mean average precision(mAP), recall, precision, F1 and other performance indicators were used to evaluate and compare the two methods. Results When Yolov5 was adopted, them AP, recall, precision and F1 were 87.56%, 82.05%, 89.44% and 0.845, respectively. When Yolov5 and Resnet34 were used in combination, the recall rate, precision rate and F1 reached 88.41%, 88.68% and 0.885, respectively. Conclusion The Yolov5 network model can help diagnose lumbar MCs. When Yolov5 and Resnet 34 models are used together, the performance is improved, which indicates that it is possible to use the Yolo series network in the diagnosis of spine diseases.

Key words: Modic changes, Magnetic resonance imaging, End-plate osteochondritis, Yolov5, Resnet34

中图分类号:

R681.5

王磊,张帅,刘钢,由胜男,王植,朱珊,陈超,马信龙,杨强. MRI诊断140例腰椎智能网络自动检测分型MCs方法的比较[J]. 山东大学学报 (医学版), 2023, 61(3): 71-79.

WANG Lei, ZHANG Shuai, LIU Gang, YOU Shengnan, WANG Zhi, ZHU Shan, CHEN Chao, MA Xinlong, YANG Qiang. Comparison of MRI diagnosis of 140 cases of MCs using intelligent network automatic detection and classification methods[J]. Journal of Shandong University (Health Sciences), 2023, 61(3): 71-79.

参考文献

[1] Hartvigsen J, Hancock MJ, Kongsted A, et al. What low back pain is and why we need to pay attention[J]. Lancet, 2018, 391(10137): 2356-2367.
[2] GBD 2013 DALYs and HALE Collaborators, Murray CJ, Barber RM, et al. Global, regional, and national disability-adjusted life years(DALYs)for 306 diseases and injuries and healthy life expectancy(HALE)for 188 countries, 1990-2013: quantifying the epidemiological transition[J]. Lancet, 2015, 386(10009): 2145-2191.
[3] Shmagel A, Foloy R, Ibrahim H, et al. Epidemiology of Chronic Low Back Pain in US Adults: Data From the 2009-2010 National health and Nutrition Examination Survey[J]. Arthrit Care Res, 2016, 68(11): 1688-1694.
[4] Karppinen J, Shen FH, Luk KD, et al. Management of degenerative disk disease and chronic low back pain[J]. Orthop Clin North Am, 2011, 42(4): 513-528.
[5] Samartzis D, Borthakur A, Belfer I, et al. Novel diagnostic and therapeutic methods for disc degeneration and low back pain[J]. Spine J, 2015, 15(9): 1919-1932.
[6] Wu A, March L, Zheng X, et al. Global low back pain prevalence and years lived with disability from 1990 to 2017: estimates from the Global Burden of Disease Study 2017[J]. Ann Transl Med, 2020, 8(6): 299.
[7] Liu J, Hao L, Suyou L, et al. Biomechanical properties of lumbar endplates and their correlation with MRI findings of lumbar degeneration[J]. J Biomech, 2016, 49(4): 586-593.
[8] Brinjikji W, Diehn FE, Jarvik JG, et al. MRI findings of disc degeneration are more prevalent in adults with low back pain than in asymptomatic controls: a systematic review and meta-analysis[J]. AJNR Am J Neuroradiol, 2015, 36(12): 2394-2399.
[9] Weishaupt D, Zanetti M, Hodler J, et al. Painful lumbar disk derangement: relevance of endplate abnormalities at MR imaging[J]. Radiology, 2001, 218(2): 420-427.
[10] Kjaer P, Korsholm L, Bendix T, et al. Modic changes and their associations with clinical findings[J]. Eur Spine J, 2006, 15(9): 1312-1319.
[11] Modic MT, Steinberg PM, Ross JS, et al. Degenerative disk disease: assessment of changes in vertebral body marrow with MR imaging[J]. Radiology, 1988, 166(pt 1): 193-199.
[12] Karchevsky M, Schweitzer ME, Carrino JA, et al. Reactive endplate marrow changes: a systematic morphologic and epidemiologic evaluation[J]. Skeletal Radiol, 2005, 34(3): 125-129.
[13] Mera Y, Teraguchi M, Hashizume H, et al. Association between types of Modic changes in the lumbar region and low back pain in a large cohort: the Wakayama spine study[J]. Eur Spine J, 2021, 30(4): 1011-1017.
[14] Yang X, Karis DSA, Vleggeert-Lankamp CLA. Association between Modic changes, disc degeneration, and neck pain in the cervical spine: a systematic review of literature[J]. Spine J, 2020, 20(5): 754-764.
[15] Jensen TS, Karppinen J, Sorensen JS, et al. Vertebral endplate signal changes(Modic change): a systematic literature review of prevalence and association with non-specific low back pain[J]. Eur Spine J, 2008, 17(11): 1407-1422.
[16] Conger A, Schuster NM, Cheng DS, et al. The effectiveness of intraosseous basivertebral nerve radiofrequency neurotomy for the treatment of chronic low back pain in patients with modic changes: a systematic review[J]. Pain Med, 2021, 22(5): 1039-1054.
[17] Modic MT, Masaryk TJ, Ross JS, et al. Imaging of degenerative disk disease[J]. Radiology, 1988, 168(1): 177-186.
[18] Setio AAA, Traverso A, de Bel T, et al. Validation, comparison, and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images: the LUNA16 challenge[J]. Med Image Anal, 2017, 42: 1-13. doi:10.1016/j.media.2017.06.015.
[19] Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis[J]. Med Image Anal, 2017, 42: 60-88. doi:10.1016/j.media.2017.07.005.
[20] Ye JJ. Artificial intelligence for pathologists is not near: it is here: description of a prototype that can transform how we practice pathology tomorrow[J]. Arch Pathol Lab Med, 2015, 139(7): 929-935.
[21] Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks[J]. Nature, 2017, 542(7639): 115-118.
[22] Gulshan V, Peng L, Coram M, et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs[J]. JAMA, 2016, 316(22): 2402-2410.
[23] GitHub. YOLOV5-Master [EB/OL]. [2022-03-01]. Available online. https://github.com/ultralytics/yolov5.git/(accessed on 1 March 2022).
[24] Yang L, Hanneke S, Carbonell J. A theory of transfer learning with applications to active learning[J]. Mach Learn, 2013, 90(2): 161-189.
[25] Rosa G, Papa J, Marana A, et al. Fine-tuning convolutional neural networks using harmony search[C] //Prog Pattern Recognit Image Anal Comput Vis Appl, 2015: 683-690. doi:10.1007/978-3-319-25751-8_82.
[26] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition [C] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016:770-778. http://image-net.org/challenges/LSVRC/2015/, http://mscoco.org/dataset/#detections-challenge2015.
[27] Jamaludin A, Lootus M, Kadir T, et al. Issls prize in bioengineering science 2017: automation of reading of radiological features from magnetic resonance images(mris)of the lumbar spine without human intervention is comparable with an expert radiologist[J]. Eur Spine J, 2017, 26(5): 1374-1383.
[28] Gao KT, Tibrewala R, Hess M, et al. Automatic detection and voxel-wise mapping of lumbar spine Modic changes with deep learning[J]. JOR Spine, 2022, 5(2): e1204.
[29] Windsor R, Jamaludin A, Kadir T, et al. SpineNetV2: Automated detection, labelling and radiological grading of clinical mr scans[J]. EESS, 2022. doi:10.48550/arXiv.2205.01683.

相关文章 15

[1]	杨咏青,赵鹏,汪玉,马文静,田迷迷,程亚旎,祖璐,林祥涛. 细胞外容积分数对62例不同病理类型肺癌的诊断价值[J]. 山东大学学报 (医学版), 2023, 61(2): 88-94.
[2]	赵恩举,赵硕,郭云亮,王锡明. 282例颈动脉钙化与脑小血管病MRI总负荷评分的关联性[J]. 山东大学学报 (医学版), 2023, 61(1): 38-44.
[3]	陶国伟,王芳,董向毅,徐亚瑄,赵琳丽,胡蓓蓓. 子宫腺肌病的超声与MRI诊断及进展[J]. 山东大学学报 (医学版), 2022, 60(7): 56-65.
[4]	左立平,蒋丰洋,周斌彬,范金蕾,梁永锋,邓展昊,于德新. 术前MRI在预测169例肝细胞肝癌微血管侵犯及早期复发的价值[J]. 山东大学学报 (医学版), 2022, 60(3): 89-95.
[5]	冯宝民,王舟,徐晗,李佳存,于乔文,修建军. 抗髓鞘少突胶质细胞糖蛋白IgG抗体相关疾病临床及影像特征[J]. 山东大学学报 (医学版), 2022, 60(3): 45-50.
[6]	刘学业,李齐明,唐弘毅,徐秋平,陈文倩,郭泾. 年轻成人颞下颌关节髁突体积、表面积与关节盘矢向位置的关系[J]. 山东大学学报 (医学版), 2021, 59(6): 117-121.
[7]	张高瑞,张玉婷,赵雨萱,王方青,于德新. MnFe₂O₄@CNS纳米探针在胰腺癌诊疗一体化中的价值[J]. 山东大学学报 (医学版), 2021, 59(4): 48-55.
[8]	宋珍珍,孙小玲,李海鸥,王芳,张冉,于德新. 120例胶质瘤及瘤周水肿MRI影像组学在评估肿瘤复发中的价值[J]. 山东大学学报 (医学版), 2021, 59(11): 53-60.
[9]	陈晓丽,桂振朝,高杨,邢梦瑶,修建军. 13例涎腺导管癌的影像学表现分析[J]. 山东大学学报 (医学版), 2021, 59(1): 78-82.
[10]	孙珊珊,房雷,赵翠萍,钟庆,高翔,李玲. 伴可逆性后部脑病综合征MR表现的急性间歇性卟啉病1例报告并文献复习[J]. 山东大学学报 (医学版), 2020, 58(2): 118-121.
[11]	陆菁菁,夏宇. 子宫内膜异位症的影像学诊断[J]. 山东大学学报 (医学版), 2019, 57(6): 40-45.
[12]	张晓倩,孟祥水,任庆国,南晓敏,安盼盼,帅欣艳,夏晓娜,王璇. 磁共振波谱成像对检测非痴呆型血管性认知障碍的探讨[J]. 山东大学学报 (医学版), 2019, 57(4): 42-46.
[13]	张丽红,王林省,陈东风,陈月芹,李娴,刘艳杰,李磊. 肾脏混合性上皮间质瘤的CT和MRI表现[J]. 山东大学学报 (医学版), 2018, 56(7): 70-75.
[14]	张丽红,李娴,王林省,李宏磊,李磊. 含脂肪节细胞神经瘤的影像学表现与病理对照[J]. 山东大学学报 (医学版), 2018, 56(12): 73-78.
[15]	王玉红,张丽红,王林省,陈月芹,王彦辉,王皆欢,李传福. 消化道颗粒细胞瘤的影像学表现[J]. 山东大学学报（医学版）, 2017, 55(8): 66-70.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed