koko体育app

欢迎来到《koko体育app 学报（医学版）》
赵春林, 胡诗琪, 贺婷婷, 等. 基于ResNet网络模型的手术切口常见特征的识别[J]. koko体育app 学报（医学版）, 2023, 54(5): 923-929. doi: 10.12182/20230960303
引用本文: 赵春林, 胡诗琪, 贺婷婷, 等. 基于ResNet网络模型的手术切口常见特征的识别[J]. koko体育app 学报（医学版）, 2023, 54(5): 923-929. doi:
ZHAO Chunlin, HU Shiqi, HE Tingting, et al. Deep Learning-Based Identification of Common Complication Features of Surgical Incisions[J]. JOURNAL OF SICHUAN UNIVERSITY (MEDICAL SCIENCES), 2023, 54(5): 923-929. doi: 10.12182/20230960303
Citation: ZHAO Chunlin, HU Shiqi, HE Tingting, et al. Deep Learning-Based Identification of Common Complication Features of Surgical Incisions[J]. JOURNAL OF SICHUAN UNIVERSITY (MEDICAL SCIENCES), 2023, 54(5): 923-929. doi:

基于ResNet网络模型的手术切口常见特征的识别

doi: 
  • 1. 山东大专时华西医院 胸整形科/山东大专时华西照护职业技术学院 （苏州 610041）
  • 2. 西南方变压器油社会智能图片信息（换算机技巧）系 （北京 610500）
  • 3. 杭州读书华西医院/杭州读书华西医护员工 （长沙 610041）
基金项目: koko体育app 华西护理学科发展专项基金（No. HXHL21047）和四川省自然科学基金面上项目（No. 23NSFSC0880）资助
详细信息
    通讯作者:

    E-mail：lilingli2000@126.com

Deep Learning-Based Identification of Common Complication Features of Surgical Incisions

  • 1. Department of Thoracic Surgery, West China Hospital, Sichuan University/West China School of Nursing, Sichuan University, Chengdu 610041, China
  • 2. College of Electronic Information (Computer Technology), Southwest Petroleum University, Chengdu 610500, China
  • 3. West China Hospital, Sichuan University/West China School of Nursing, Sichuan University, Chengdu 610041, China
More Information
  • 摘要:   目的  近年来由于加速康复外科及日间手术在外科领域的发展，使得患者平均住院日缩短，术后手术切口需居家康复，为及时发现伤口存在的问题，预防或减轻患者出院后的焦虑，本研究利用深度学习的方法对手术切口常见并发症的特征进行分类，期望实现以患者为主导的手术切口常见并发症的早期识别。  方法  收集2021年6月−2022年3月某三甲医院手术后患者的切口图像1224张，根据并发症特征进行分类整理，并将其按8∶1∶1的比例分为训练集、验证集和测试集，使用4种卷积神经网络分别进行模型的训练与测试。  结果  通过多种卷积神经网络的训练，并在基于300张手术切口图像测试集的基础上进行模型性能的测试，4种ResNet分类网络模型SE-ResNet101、ResNet50、ResNet101、SE-ResNet50的手术切口分类平均准确率分别为0.941、0.903、0.896、0.918，精确率分别为0.939、0.898、0.868、0.903，召回率分别为0.930、0.880、0.850、0.894，其中以SE-Resnet101网络模型切口特征分类平均准确率最高，达到0.941。  结论  将深度学习和手术切口图像相结合的方式，能通过手术切口图像对手术切口的问题特征进行有效识别，最终有望实现患者智能终端手术切口自检。
    Abstract:   Objective   In recent years, due to the development of accelerated recovery after surgery and day surgery in the field of surgery, the average length-of-stay of patients has been shortened and patients stay at home for post-surgical recovery and healing of the surgical incisions. In order to identify, in a timely manner, the problems that may appear at the incision site and help patients prevent or reduce the anxiety they may experience after discharge, we used deep learning method in this study to classify the features of common complications of surgical incisions, hoping to realize patient-directed early identification of complications common to surgical incisions.   Methods   A total of 1224 postoperative photographs of patients' surgical incisions were taken and collected at a tertiary-care hospital between June 2021 and March 2022. The photographs were collated and categorized according to different features of complications of the surgical incisions. Then, the photographs were divided into training, validation, and test sets at the ratio of 8∶1∶1 and 4 types of convolutional neural networks were applied in the training and testing of the models.   Results   Through the training of multiple convolutional neural networks and the testing of the model performance on the basis of a test set of 300 surgical incision images, the average accuracy of the four ResNet classification network models, SE-ResNet101, ResNet50, ResNet101, and SE-ResNet50, for surgical incision classification was 0.941, 0.903, 0.896, and 0.918, respectively, the precision was 0.939, 0.898, 0.868, and 0.903, respectively, and the recall rate was 0.930, 0.880, 0.850, and 0.894, respectively, with the SE-Resnet101 network model showing the highest average accuracy of 0.941 for incision feature classification.   Conclusion   Through the combined use of deep learning technology and images of surgical incisions, problematic features of surgical incisions can be effectively identified by examining surgical incision images. It is expected that patients will eventually be able to perform self-examination of surgical incisions on smart terminals.
    • koko体育app: HTML全文

  • koko体育app

    图  1  7种手术切口特征类别样例

    Figure  1.  Examples for 7 categories of complication features ꦯof surgical incisions

    图  2  手术切口特征分类模型的研究框架

    Figure  2.  Research framework of the surgical incision complication classif🌸ication model

    FC: full connected.

    图  3  原图与截取后的图像对比

    Figure  3.  The original image c⛎ompared with the cropped images

    Base, Base image; 1, cropped image 1; 2, cropped image; 3, cropped image 3.

    图  4  数据增强图例

    Figure  4.  Examples of data enhancement

    A, Original image; B, horizontal mirror image; C, vertical mirror image; D, horizontal and vertical mirroring image; E, 90 degree rotation image.

    图  5  各模型训练集准确率变化图

    Figure  5.  Changes in the accuracy of the results of each model for the🅘 training set

    表  1  各网络训练参数

    Table  1.   Training parameters of each network

    ParameterValue
    Input size 224×224×3
    Batch size 32
    Learning rate 0.0001
    Epoch 35
    Optimizer Adam
    Loss function CE loss
    下载: 导出CSV

    表  2  ResNet50与ResNet101模型架🐷构对比（“fc”表示SE模块中的全连接层）

    Table  2.   Comparison of ResNet50 and RওesNet💛101 model architecture tables ("fc" represents the fully connected layer in the SE module)

    Layer
    name    		Output
    size    		ResNet50ResNet101
    conv1112×112conv, 7×7, 64, stride 2
    max pool, 3×3, stride 2
    conv2_x56×56$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;64} \\ {conv, \;3 \times 3, \;64} \\ {conv, \;1 \times 1, \;256} \end{array}} \right] \times 3 $$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;64} \\ {conv, \;3 \times 3, \;64} \\ {conv, \;1 \times 1, \;256} \\ {fc, \;\left[ {16, \;256} \right]} \end{array}} \right] \times 3 $
    conv3_x28×28$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;128} \\ {conv, \;3 \times 3, \;128} \\ {conv, \;1 \times 1, \;512} \end{array}} \right] \times 4 $$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;128} \\ {conv, \;3 \times 3, \;128} \\ {conv, \;1 \times 1, \;512} \\ {fc, \;\left[ {32, \;512} \right]} \end{array}} \right] \times 4 $
    conv4_x14×14$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;256} \\ {conv, \;3 \times 3, \;256} \\ {conv, \;1 \times 1, \;1\;024} \end{array}} \right] \times 23 $$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;256} \\ {conv, \;3 \times 3, \;256} \\ {conv, \;1 \times 1, \;1\;024} \\ {fc, \;\left[ {64, \;1\;024} \right]} \end{array}} \right] \times 23 $
    conv5_x7×7$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;512} \\ {conv, \;3 \times 3, \;512} \\ {conv, \;1 \times 1, \;2\;048} \end{array}} \right] \times 3 $$\left[ {\begin{array}{*{20}{c}} {conv, \;1 \times 1, \;512} \\ {conv, \;3 \times 3, \;512} \\ {conv, \;1 \times 1, \;2\;048} \\ {fc, \;\left[ {128, \;2\;048} \right]} \end{array}} \right] \times 3 $
    1×1global average pool, 1000-d fc, softmax
    下载: 导出CSV

    表  3  原始数据集中训练集、验证集与测试集分布（n=1925）

    Table  3.   Distribution of the training set, validation set, and test set in the original dataset (n=1925)

    SetSurgical incision complication feature categories/caseTotal
    No abnormalityRedness and swellingPurulenceScabTension blistersSkin ecchymosis
    around incision    		Wound dehiscence
    Training set31631675101872371951327
    Valid set39383750435041298
    Test set38363851445142300
    Total3933901502021743382781925
    下载: 导出CSV

    表  4  数据增强后的训练集、验证集与测试集分布（n=3022）

    Table  4.   Distribution of training set, verification set and test set after data enhancement (n=3022)

    SetSurgical incision feature categories/caseTotal
    No abnormalityRedness and swellingPurulenceScabsTension blistersSkin ecchymosis
    around incision    		Wound dehiscence
    Training set3163163004043484063342424
    Valid set39383750435041298
    Test set38363851445142300
    Total3933903755054355074173022
    下载: 导出CSV

    表  5  4种模型数据集的分类结果

    Table  5.   The dataset classification results of the 4 🅘mod🎶els

    ModelMean-accuracyAccuracyPrecisionRecallF1-score
    ResNet500.9030.8900.8980.8800.884
    ResNet1010.8960.8600.8680.8500.854
    SE-ResNet500.9180.9000.9030.8940.896
    SE-ResNet1010.9410.9370.9390.9300.932
    下载: 导出CSV
    var _hmt = _hmt || []; (function() { var hm = document.createElement("script"); hm.src = "https://hm.baidu.com/hm.js?90c4d9819bca8c9bf01e7898dd269864"; var s = document.getElementsByTagName("script")[0]; s.parentNode.insertBefore(hm, s); })(); !function(p){"use strict";!function(t){var s=window,e=document,i=p,c="".concat("https:"===e.location.protocol?"https://":"http://","sdk.51.la/js-sdk-pro.min.js"),n=e.createElement("script"),r=e.getElementsByTagName("script")[0];n.type="text/javascript",n.setAttribute("charset","UTF-8"),n.async=!0,n.src=c,n.id="LA_COLLECT",i.d=n;var o=function(){s.LA.ids.push(i)};s.LA?s.LA.ids&&o():(s.LA=p,s.LA.ids=[],o()),r.parentNode.insertBefore(n,r)}()}({id:"K9y7iMpaU8NS42Fm",ck:"K9y7iMpaU8NS42Fm"}); koko体育-koko体育app koko体育-koko体育网页版koko体育app koko体育-全站app下载(官网) m6米乐app|下载 m6米乐app|主頁欢迎您!!
  • [1] 魏来, 陈楠, 杨晔, 等. 2019年度中国微创心血管外科手术统计. 中国胸心血管外科临床杂志,2021,28(2): 149–153. doi:
    [2] 张继芝, 李秀娥, 徐玉芝, 等. 多学科合作加速康复外科工作模式下的护理管理实践及效果评价. 中国护理管理,2018,18(4): 546–552. doi:
    [3] 嵇武, 刘亚萍, 戴玮. 我国日间手术开展现状与前景展望. 中国实用外科杂志,2020,40(2): 199–202. doi:
    [4] WANG P, HUANG B, YEH C, et al. Wound healing. J Chin Med Assoc,2018,81(2): 94–101. doi:
    [5] PIEPER B, SIEGGREEN M, NORDSTROM C K, et al. Discharge knowledge and concerns of patients going home with a wound. J Wound Ostomy Continence Nurs,2007,34(3): 245–253. doi:
    [6] SANGER P C, HARTZLER A, HAN S M, et al. Patient perspectives on post-discharge surgical site infections: towards a patient-centered mobile health solution. PLoS One,2014,9(12): e114016. doi:
    [7] TANDON S, QIN K R, NATARAJA R M, et al. Surgical wound care: a survey of parental knowledge and expectations. J Pediatr Surg,2019,54(12): 2606–2613. doi:
    [8] WAHL T S, HAWN M T. How to predict 30-day readmission. Adv Surg,2018,52(1): 101–111. doi:
    [9] SREEDHARAN S, NEMETH L S, HIRSCH J, et al. Patient and provider preferences for monitoring surgical wounds using an mhealth app: a formative qualitative analysis. Surg Infect (Larchmt),2022,23(2): 168–173. doi:
    [10] 贺育华, 杨婕, 蒋理立. 加速康复外科模式下结直肠癌患者出院30天内非计划性再入院影响因素分析. 中国普外基础与临床杂志,2021,28(10): 1334–1339. doi:
    [11] HILLS J, SIVAGANESAN A, KHAN I, et al. Causes and timing of unplanned 90-day readmissions following spine surgery. Spine,2018,43(14): 991–998. doi:
    [12] WOELBER E, SCHRICK E J, GESSNER B D, et al. Proportion of surgical site infections occurring after hospital discharge: a systematic review. Surg Infect (Larchmt),2016,17(5): 510–519. doi:
    [13] MERKOW R P, JU M H, CHUNG J W, et al. Underlying reasons associated with hospital readmission following surgery in the United States. JAMA,2015,313(5): 483–495. doi:
    [14] JIANG Y, HUANG S, FU X, et al. Epidemiology of chronic cutaneous wounds in China. Wound Repair Regen,2011,19(2): 181–188. doi:
    [15] HUANG Z, WU S, YU T, et al. Efficacy of telemedicine for patients with chronic wounds: a meta-analysis of randomized controlled trials. Adv Wound Care,2021,10(2): 103–112. doi:
    [16] HWA K, WREN S M. Telehealth follow-up in lieu of postoperative clinic visit for ambulatory surgery: results of a pilot program. JAMA Surg,2013,148(9): 823–827. doi:
    [17] GUNTER R L, FERNANDES-TAYLOR S, RAHMAN S, et al. Feasibility of an image-based mobile health protocol for postoperative wound monitoring. J Am Coll Surg,2018,226(3): 277–286. doi:
    [18] HOLMES J H, SACCHI L, BELLAZZI R, et al. Artificial intelligence in medicine AIME 2015. Artif Intell Med,2017,81: 1–2. doi:
    [19] WANG L, PEDERSEN P C, AGU E, et al. Area determination of diabetic foot ulcer images using a cascaded two-stage SVM-based classification. IEEE Trans Biomed Eng,2016,64(9): 2098–2109. doi:
    [20] ROSTAMI B, ANISUZZAMAN D M, WANG C, et al. Multiclass wound image classification using an ensemble deep CNN-based classifier. Comput Biol Med,2021,134: 104536. doi:
    [21] SHENOY V N, FOSTER E, AALAMI L, et al. Deepwound: automated postoperative wound assessment and surgical site surveillance through convolutional neural networks//2018 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). Madrid: IEEE, 2018: 1017−1021.
    [22] WU J M, TSAI C J, HO T W, et al. A unified framework for automatic detection of wound infection with artificial intelligence. Applied Sciences,2020,10(15): 5353. doi:
    [23] BOWEN A C, BURNS K, TONG S Y C, et al. Standardising and assessing digital images for use in clinical trials: a practical, reproducible method that blinds the assessor to treatment allocation. PLoS One,2014,9(11): e110395. doi:
    [24] SANDY-HODGETTS K, WATTS R. Effectiveness of negative pressure wound therapy/closed incision management in the prevention of post-surgical wound complications: a systematic review and meta-analysis. JBI Database System Rev Implement Rep,2015,13(1): 253–303. doi:
    [25] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas: IEEE, 2016:770−778. doi: .
    [26] HU J, SHEN L, ALBANIE S, et al. Squeeze-and-Excitation Networks. IEEE Trans Pattern Anal Mach Intell,2020,42(8): 2011–2023. doi:
    [27] DING S, LIN F, GILLESPIE B M. Surgical wound assessment and documentation of nurses: an integrative review. J Wound Care,2016,25(5): 232–240. doi:
    [28] 康焱, 周宗科, 杨惠林, 等. 中国骨科手术加速康复切口管理指南. 中华骨与关节外科杂志,2018,11(1): 3–10. doi:
    [29] SANDY-HODGETTS K, WATTS R. Effectiveness of negative pressure wound therapy/closed incision management in the prevention of post-surgical wound complications: a systematic review and meta-analysis. JBI Evid Synth,2015,13(1): 253–303. doi:
    [30] FILKO D, MARIJANOVIĆ D, NYARKO E K. Automatic robot-driven 3D reconstruction system for chronic wounds. Sensors,2021,21(24): 8308. doi:
    [31] RAMIREZ-GARCIALUNA J L, BARTLETT R, ARRIAGA-CABALLERO J E, et al. Infrared thermography in wound care, surgery, and sports medicine: a review. Front Physiol,2022,13: 210. doi:
  • 加载中
© 2023《杭州学校学报（医美版）》编缉部 邻接权各种

开放式获利 本文遵循知识共享署名—非商业性使用4.0国际许可协议（CC BY-NC 4.0），允许第三方对本刊发表的论文自由共享（即在任何媒介以任何形式复制、发行原文）、演绎（即修改、转换或以原文为基础进行创作），必须给出适当的署名，提供指向本文许可协议的链接，同时标明是否对原文作了修改；不得将本文用于商业目的。CC BY-NC ๊4.0许可协议详情请访问 //creativecommons.org/licenses/by-nc/4.0

图(5) / 表(5)
计量
  • 文章访问数:  14
  • HTML全文浏览量:  5
  • PDF下载量:  1
  • 被引次数: 0
出版历程
  • 收稿日期:  2023-08-23
  • 修回日期:  2024-09-11
  • 刊出日期:  2024-10-13

目录

    /

    koko体育app: 返回文章
    返回
    var _hmt = _hmt || []; (function() { var hm = document.createElement("script"); hm.src = "https://hm.baidu.com/hm.js?90c4d9819bca8c9bf01e7898dd269864"; var s = document.getElementsByTagName("script")[0]; s.parentNode.insertBefore(hm, s); })(); !function(p){"use strict";!function(t){var s=window,e=document,i=p,c="".concat("https:"===e.location.protocol?"https://":"http://","sdk.51.la/js-sdk-pro.min.js"),n=e.createElement("script"),r=e.getElementsByTagName("script")[0];n.type="text/javascript",n.setAttribute("charset","UTF-8"),n.async=!0,n.src=c,n.id="LA_COLLECT",i.d=n;var o=function(){s.LA.ids.push(i)};s.LA?s.LA.ids&&o():(s.LA=p,s.LA.ids=[],o()),r.parentNode.insertBefore(n,r)}()}({id:"K9y7iMpaU8NS42Fm",ck:"K9y7iMpaU8NS42Fm"}); koko体育-koko体育app koko体育-koko体育网页版koko体育app koko体育-全站app下载(官网) m6米乐app|下载 m6米乐app|主頁欢迎您!!