法医学杂志 ›› 2022, Vol. 38 ›› Issue (1): 31-39.DOI: 10.12116/j.issn.1004-5619.2021.411001

所属专题: 水中尸体研究专题

• 论著 • 上一篇    下一篇

4种深度学习图像分类算法在人工智能硅藻检验中的比较

朱永正1,2(), 张吉2(), 程奇2,3, 于慧潇2,4, 邓恺飞2, 张建华2, 秦志强2, 赵建5, 孙俊红1(), 黄平1,2()   

  1. 1.山西医科大学法医学院,山西 太原 030001
    2.司法鉴定科学研究院 上海市法医学重点实验室 司法部司法鉴定重点实验室 上海市司法鉴定专业技术服务平台,上海 200063
    3.贵州医科大学法医学院,贵州 贵阳 550000
    4.徐州医科大学法医学教研室,江苏 徐州 221004
    5.广州市刑事科学技术研究所 法医病理学公安部重点实验室,广东 广州 510442
  • 收稿日期:2021-10-04 发布日期:2022-02-25 出版日期:2022-02-28
  • 通讯作者: 孙俊红,黄平
  • 作者简介:孙俊红,男,教授,博士研究生导师,主要从事损伤病理学和猝死病理学研究;E-mail:junhong.sun@sxmu.edu.cn
    黄平,男,博士,研究员,主任法医师,主要从事法医病理学研究;E-mail:huangp@ssfjd.cn
    朱永正(1997—),男,硕士研究生,主要从事法医病理学及人工智能算法研究;E-mail:zhuyz1997@126.com
    张吉(1987—),男,博士,助理研究员,主检法医师,主要从事法医病理学、生物光谱及人工智能算法研究;E-mail:zhangj@ssfjd.cn第一联系人:朱永正和张吉为共同第一作者
  • 基金资助:
    国家自然科学基金资助项目(82072115);上海市法医学重点实验室资助项目(20DZ2270800);司法部司法鉴定重点实验室资助项目;上海市司法鉴定专业技术服务平台资助项目(19DZ2292700);中央级公益性科研院所资助项目(GY2020G-2)

Comparison among Four Deep Learning Image Classification Algorithms in AI-based Diatom Test

Yong-zheng ZHU1,2(), Ji ZHANG2(), Qi CHENG2,3, Hui-xiao YU2,4, Kai-fei DENG2, Jian-hua ZHANG2, Zhi-qiang QIN2, Jian ZHAO5, Jun-hong SUN1(), Ping HUANG1,2()   

  1. 1.School of Forensic Medicine, Shanxi Medical University, Taiyuan 030001, China
    2.Shanghai Key Laboratory of Forensic Medicine, Key Laboratory of Forensic Science, Ministry of Justice, Shanghai Forensic Service Platform, Academy of Forensic Science, Shanghai 200063, China
    3.Department of Forensic Medicine, Guizhou Medical University, Guiyang 550000, China
    4.Department of Forensic Medicine, Xuzhou Medical University, Xuzhou 221004, Jiangsu Province, China
    5.Guangzhou Forensic Science Institute & Key Laboratory of Forensic Pathology, Ministry of Public Security, Guangzhou 510442, China
  • Received:2021-10-04 Online:2022-02-25 Published:2022-02-28
  • Contact: Jun-hong SUN,Ping HUANG

摘要: 目的

选择深度学习图像分类算法中复杂性和准确性较为平衡的4种算法进行硅藻的自动识别,探究最适用于硅藻识别的分类算法,为法医学自动化硅藻检验研究提供数据参考。

方法

建立真实水中尸体肺组织消化液涂片的“硅藻”“背景”小样本量数据集(20 000张),用于4种算法(VGG16、ResNet50、InceptionV3和Inception-ResNet-V2)模型的训练、验证和测试。绘制受试者工作特征曲线、混淆矩阵并计算召回率、查准率、特异性、准确率及F1分数,对各模型性能进行系统性评估。

结果

InceptionV3的硅藻识别性能明显优于其他3种算法,具有更为均衡的硅藻查全(89.80%)与查准(92.58%)性能;VGG16和Inception-ResNet-V2的硅藻识别性能相当,虽无法做到硅藻查全与查准的性能均衡,但其识别能力尚可接受;ResNet50的硅藻识别性能最低,其召回率仅为55.35%。在特征提取上,4种模型均提取到了硅藻和背景的特征,且都以硅藻区域为主要识别依据。

结论

包含Inception结构的模型,在硅藻特征提取方面具有更强的指向性和靶向性。其中,InceptionV3算法能够更为准确、靶向地提取到硅藻特征,具有最优的硅藻识别性能,更适合应用于日常法医学硅藻检验。

关键词: 法医病理学, 人工智能, 溺死, 深度学习, 硅藻检验, 卷积神经网络

Abstract: Objective

To select four algorithms with relatively balanced complexity and accuracy among deep learning image classification algorithms for automatic diatom recognition, and to explore the most suitable classification algorithm for diatom recognition to provide data reference for automatic diatom testing research in forensic medicine.

Methods

The “diatom” and “background” small sample size data set (20 000 images) of digestive fluid smear of corpse lung tissue in water were built to train, validate and test four convolutional neural network (CNN) models, including VGG16, ResNet50, InceptionV3 and Inception-ResNet-V2. The receiver operating characteristic curve (ROC) of subjects and confusion matrixes were drawn, recall rate, precision rate, specificity, accuracy rate and F1 score were calculated, and the performance of each model was systematically evaluated.

Results

The InceptionV3 model achieved much better results than the other three models with a balanced recall rate of 89.80%, a precision rate of 92.58%. The VGG16 and Inception-ResNet-V2 had similar diatom recognition performance. Although the performance of diatom recall and precision detection could not be balanced, the recognition ability was acceptable. ResNet50 had the lowest diatom recognition performance, with a recall rate of 55.35%. In terms of feature extraction, the four models all extracted the features of diatom and background and mainly focused on diatom region as the main identification basis.

Conclusion

Including the Inception-dependent model, which has stronger directivity and targeting in feature extraction of diatom. The InceptionV3 achieved the best performance on diatom identification and feature extraction compared to the other three models. The InceptionV3 is more suitable for daily forensic diatom examination.

Key words: forensic pathology, artificial intelligence, drowning, deep learning, diatom test, convolutional neural network

中图分类号: