基于改进YOLOv11的CT图像中的脑膜瘤检测OA
Meningioma detection in CT images based on improved YOLOv11
针对脑膜瘤检测任务中存在的形状和尺度不一、背景复杂等问题,提出一种基于改进YOLOv11(you only look once第11版)的CT图像中的脑膜瘤检测算法.首先,通过设计多维注意力融合机制并引入深度可分离卷积优化网络架构.其次,使用DIoU作为边界框损失函数加快模型收敛速度.最后,通过具体实验验证所提算法的有效性,并与其他算法进行对比研究.结果表明,所提算法的准确度、召回率和平均精度值分别达到96.2％,99.1％和98.8％,与其他5种算法相比,均为最高.
To address challenges such as inconsistent shapes and sizes,as well as complex backgrounds in meningioma detection tasks,this paper proposed a meningioma detection algorithm for CT images based on an improved YOLOv11(You Only Look Once version 11).First,designing a multi-dimensional attention fusion mechanism and introducing a depthwise separable convolution optimize the network architecture.Second,using DIoU as the bounding box loss function accelerates model convergence.Finally,the effectiveness of the proposed method was validated through specific experiments and compared with other algorithms.The results indicate that compared with the other five algorithms,the proposed algorithm achieved the highest mean values for accuracy,recall,and average precision,reaching 96.2％,99.1％,and 98.8％,respectively.
江永成;刘元志;胡根生;魏子靖
安徽大学电气工程与自动化学院,安徽 合肥 230601安徽大学电气工程与自动化学院,安徽 合肥 230601安徽大学电子信息工程学院,安徽 合肥 230601安徽大学电气工程与自动化学院,安徽 合肥 230601
信息技术与安全科学
脑膜瘤检测YOLOv11深度可分离卷积注意力机制损失函数
meningioma detectionYOLOv11depthwise separable convolutionattention mechanismloss function
《安徽大学学报（自然科学版）》 2026 (3)
53-59,7
国家自然科学基金资助项目(32372632,52175210)
评论