融合双注意力机制的骨质疏松分类算法OA
A Dual-attention Mechanism Fusion Algorithm for Osteoporosis Classification
提出了一种融合双注意力机制的骨质疏松分类算法(R-ENet),该方法基于EfficientNetV2网络的思想,在原EfficientNetV2-small网络结构的基础上,用分组策略去优化原始的网络层结构,既保证了轻量化计算,又增强了模型对细粒度特征的辨识力,并对原始分类头结构进行改进,提出一种新的分类头结构,进一步提升模型的特征判别能力,提高分类准确率,最终将膝关节的X射线图像分类为骨质疏松症、骨质减少和正常3个类别.该研究中涉及的公开膝关节X光片数据集已被用于训练,验证和测试所提出的模型,实验结果表明,提出的模型在数据集上的分类准确率最高,达到了95.0%,相比最优对比网络提高了3.0%.提高了深度学习用于骨质疏松症的诊断准确性,为今后基于图像的疾病诊断过程提供有价值的参考.
An osteoporosis classification algorithm with a fused dual-attention mechanism,named R-ENet is proposed.The method builds upon the EfficientNetV2 architecture and optimizes the original EfficientNetV2-small network using a grouping strategy,which ensures lightweight computation while enhancing the model's ability to recognize fine-grained features.Additionally,the original classification head is improved with a novel structure to further boost feature discrimination and classification accuracy.The model classifies knee X-ray images into three categories:osteoporosis,osteopenia,and normal.A publicly available knee X-ray dataset is used to train,validate,and test the proposed model.Experimental results demonstrate that the proposed model achieves the highest classification accuracy on the dataset,reaching 95.0%,which represents a 3.0%improvement over the best-performing baseline network.This work enhances the diagnostic accuracy of deep learning in osteoporosis detection and provides a valuable reference for future image-based diagnostic processes.
王庆顺;王沫楠
东北林业大学 机电工程学院,哈尔滨 150040东北林业大学 机电工程学院,哈尔滨 150040
信息技术与安全科学
X射线图像骨质疏松双注意力机制分类器图像分类卷积神经网络
X-ray imagingosteoporosisdual-attention mechanismclassifierimage classificationconvolutional neural network
《机电工程技术》 2026 (17)
19-25,7
国家自然科学基金(61972117)黑龙江省双一流学科协同创新成果项目(LJGXCG2025-P03)
评论