| 121 | 0 | 528 |
| 下载次数 | 被引频次 | 阅读次数 |
为提高膝骨关节炎X光影像分类精度，特别是对难分类样本的识别能力，提出了一种改进的EfficientNetV2＿S模型。该模型在保留EfficientNetV2＿S轻量化特性的基础上，引入高效特征融合网络EffNet增强多尺度特征的语义表达;同时提出带阈值的自适应权重惩罚损失函数Taploss，动态调整对难易样本的惩罚权重。在骨关节炎倡议数据集上的实验表明，改进模型显著提升了分类性能，准确率达到71.4%，尤其对分类难度高的等级1图像识别效果提升明显。模型在Mendeley Data数据集上也展现出良好的泛化能力。相较于当前先进非集成模型，该模型在精度与轻量化方面具有优势。
Abstract：To improve the classification accuracy of knee osteoarthritis X-ray images,particularly for hard-to-classify samples,an improved EfficientNetV2 ＿ S model is proposed. While retaining the lightweight characteristics of EfficientNetV2＿S,the model introduces an efficient feature fusion network( EffNet) to enhance the semantic representation of multi-scale features. Additionally,a threshold-based adaptive penalty loss function( Taploss) is proposed to dynamically adjust the penalty weights for easy and hard samples. Experiments on the Osteoarthritis Initiative( OAI) dataset demonstrate that the improved model significantly enhances classification performance,achieving an accuracy of 71. 4%,with particularly notable improvements in the recognition of challenging Grade 1 images. The model also exhibits strong generalization capabilities on the Mendeley Data dataset. Compared to current state-of-the-art non-ensemble models,the proposed model demonstrates advantages in both accuracy and lightweight design.
[1] SAINI D,CHAND T,CHOUHAN D K,et al. A comparative analysis of automatic classification and grading methods for knee osteoarthritis focussing on X-ray images[J]. Biocybernetics and Biomedical Engineering,2021,41(2):419-444.
[2] KOHN M D,SASSOON A A,FERNANDO N D. Classifications in brief:Kellgren-Lawrence classification of osteoarthritis[J].Clinical Orthopaedics and Related Research,2016,474:1886-1893.
[3] ZHAO X,SHEN Z,CHEN D,et al. One-shot traumatic brain segmentation with adversarial training and uncertainty rectification[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham:Springer Nature Switzerland,2023:120-129.
[4] ANTHIMOPOULOS M,CHRISTODOULIDIS S,EBNER L,et al. Lung pattern classification for interstitial lung diseases using a deep convolutional neural network[J]. IEEE Transactions on Medical Imaging,2016,35(5):1207-1216.
[5] SKREDE O J,DE RAEDT S,KLEPPE A,et al. Deep learning for prediction of colorectal cancer outcome:a discovery and validation study[J]. The Lancet,2020,395(10221):350-360.
[6] XU Z,WANG Y,CHEN M,et al. Multi-region radiomics for artificially intelligent diagnosis of breast cancer using multimodal ultrasound[J]. Computers in Biology and Medicine,2022,149:105920.
[7] WU R,LIANG P,HUANG Y,et al. Automatic segmentation of hemorrhages in the ultra-wide field retina:multi-scale attention subtraction networks and an ultra-wide field retinal hemorrhage dataset[J/OL]. IEEE Journal of Biomedical and Health Informatics,2024:1-14.[2024-10-02]. https://doi. org/10. 1109/JBHI. 2024. 3457512.
[8] ANTONY J,MCGUINNESS K,MORAN K,et al. Automatic detection of knee joints and quantification of knee osteoarthritis severity using convolutional neural networks[C]//Machine Learning and Data Mining in Pattern Recognition:13th International Conference. Cham:Springer,2017:376-390.
[9] TIULPIN A,THEVENOT J O M,RAHTU E,et al. Automatic knee osteoarthritis diagnosis from plain radiographs:a deep learning-based approach[J]. Scientific Reports,2018,8(1):1727.
[10] GóRRIZ M,ANTONY J,MCGUINNESS K,et al. Assessing knee OA severity with CNN attention-based end-to-end architectures[C]//New York:International Conference on Medical Imaging with Deep Learning,2019:197-214.
[11] CHEN P,GAO L,SHI X,et al. Fully automatic knee osteoarthritis severity grading using deep neural networks with a novel ordinal loss[J]. Computerized Medical Imaging and Graphics,2019,75:84-92.
[12] THOMAS K A,KIDZI?SKI?，HALILAJ E,et al. Automated classification of radiographic knee osteoarthritis severity using deep neural networks[J]. Radiology:Artificial Intelligence,2020,2(2):e190065.
[13] WANG Y,BI Z,XIE Y,et al. Learning from highly confident samples for automatic knee osteoarthritis severity assessment:data from the osteoarthritis initiative[J]. IEEE Journal of Biomedical and Health Informatics,2021,26(3):1239-1250.
[14] LEE S,KIM N. Deep learning model ensemble for the accuracy of classification degenerative arthritis[J]. Computers,Materials&Continua,2023,75(1):1981-1994.
[15] REHMAN S U,GRUHN V. A sequential VGG16+CNN based automated approach with adaptive input for efficient detection of knee osteoarthritis stages[J]. IEEE Access,2024,12:62407-62415.
[16] TAN M,LE Q. Efficientnet:rethinking model scaling for convolutional neural networks[C]//New York:International Conference on Machine Learning,2019:6105-6114.
[17] TAN M,LE Q. Efficientnetv2:smaller models and faster training[C]//New York:International Conference on Machine Learning,2021:10096-10106.
[18] WANG Q,WU B,ZHU P,et al. ECA-Net:efficient channel attention for deep convolutional neural networks[C]//Piscataway:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020:11531-11539.
[19] HE K,ZHANG X,REN S,et al. Deep residual learning for image recognition[C]//Piscataway:Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016:770-778.
[20] HUANG G,LIU Z,VAN DER MAATEN L,et al. Densely connected convolutional networks[C]//Piscataway:Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2017:4700-4708.
[21] SIMONYAN K. Very deep convolutional networks for large-scale image recognition[J/OL].(2014-09-04)[2024-10-02].http://arxiv. org/abs/1409. 1556
[22] LIU Z,LIN Y,CAO Y,et al. Swin transformer:Hierarchical vision transformer using shifted windows[C]//Piscataway:Proceedings of the IEEE/CVF International Conference on Computer Vision,2021:10012-10022.
[23] LIU Z,MAO H,WU C Y,et al. A convnet for the 2020s[C]//Piscataway:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2022:11976-11986.
[24] YU W,LUO M,ZHOU P,et al. Metaformer is actually what you need for vision[C]//Piscataway:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2022:10819-10829.
[25] PENG Z,HUANG W,GU S,et al. Conformer:Local features coupling global representations for visual recognition[C]//Piscataway:Proceedings of the IEEE/CVF International Conference on Computer Vision,2021:367-376.
[26] SANDLER M,HOWARD A,ZHU M,et al. Mobilenetv2:inverted residuals and linear bottlenecks[C]//Piscataway:Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2018:4510-4520.
[27] ZHANG X,ZHOU X,LIN M,et al. Shufflenet:an extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2018:6848-6856.
[28] IANDOLA F N. Squeeze Net:Alex Net-level accuracy with 50×fewer parameters and<0. 5 MB model size[J].(2016-02-24)[2024-10-02]. http://arxiv. org/abs/1602. 07360.
[29] SELVARAJU R R,COGSWELL M,DAS A,et al. Grad-cam:visual explanations from deep networks via gradient-based localization[C]//Piscataway:Proceedings of the IEEE International Conference on Computer Vision,2017:618-626.
基本信息:
中图分类号:TP391.41;R684.3
引用信息:
[1]黄一山,刘英莉,茶寅秋,等.基于改进EfficientNetV2＿S的膝骨关节炎X光影像分类[J].陕西理工大学学报(自然科学版),2025,41(05):42-52.
基金信息:
云南省科技厅基础研究项目(202001AY070001-306)