School of Forensic Medicine, Shanxi Medical University, Taiyuan 030001, China.
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.
Fa Yi Xue Za Zhi. 2022 Feb 25;38(1):31-39. doi: 10.12116/j.issn.1004-5619.2021.411001.
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.
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 1 score were calculated, and the performance of each model was systematically evaluated.
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.
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.
从深度学习图像分类算法中选择 4 种复杂度和准确性相对均衡的算法用于自动硅藻识别，并探索最适合硅藻识别的分类算法，为法医自动硅藻检验研究提供数据参考。
构建了尸体肺组织水中消化液涂片的“硅藻”和“背景”小样本量数据集（20000 张图像），以训练、验证和测试 4 种卷积神经网络（CNN）模型，包括 VGG16、ResNet50、InceptionV3 和 Inception-ResNet-V2。绘制受试者的受试者工作特征曲线（ROC）和混淆矩阵，计算召回率、准确率、特异性、准确率和 1 分，并对每个模型的性能进行系统评估。
InceptionV3 模型的结果明显优于其他 3 种模型，其召回率为 89.80%，准确率为 92.58%。VGG16 和 Inception-ResNet-V2 具有相似的硅藻识别性能。虽然不能平衡硅藻召回率和精度检测的性能，但识别能力是可以接受的。ResNet50 的硅藻识别性能最低，召回率为 55.35%。在特征提取方面，4 种模型均提取了硅藻和背景的特征，主要以硅藻区域为主要识别依据。
包括 Inception 依赖模型，它在硅藻特征提取方面具有更强的针对性和靶向性。与其他 3 种模型相比，InceptionV3 在硅藻识别和特征提取方面表现最佳。InceptionV3 更适合日常法医硅藻检验。
