| 304 | 0 | 157 |
| 下载次数 | 被引频次 | 阅读次数 |
目的：为了提升超声图像乳腺肿瘤的检测性能，提出基于降采样（adaptive downsampling,ADown）模块和动态卷积（dynamic convolution,DynamicConv）模块的改进YOLOv11算法。方法：以原始YOLOv11算法为基础进行改进，将YOLOv11算法的主干网络（Backbone）及颈部模块（Neck）中的卷积（Conv）模块替换成ADown模块，并在C3k2的瓶颈模块（Bottleneck）引入DynamicConv模块，构建YOLOv11-ADown-DynamicConv算法。使用Kaggle平台公开的乳腺超声图像（Breast Ultrasound Image,BUSI）数据集对YOLOv11-ADown-DynamicConv算法进行训练和验证，并与原始YOLOv11算法对超声图像乳腺肿瘤的检测性能进行比较。结果：在超声图像乳腺肿瘤的检测任务中，YOLOv11-ADown-DynamicConv算法在交并比阈值为0.5时的平均精度均值mAP50与交并比阈值为0.5-0.95时的平均精度均值mAP50-95分别为0.763和0.522，浮点运算次数和每秒帧数分别为4.8G和135帧/s，均优于原始YOLOv11算法。结论：YOLOv11-ADown-DynamicConv算法在超声图像乳腺肿瘤的检测任务中表现出色，可以提升YOLOv11算法的检测性能，能够辅助医生更高效地筛查疾病、评估病情。
Abstract：Objective To propose an improved YOLOv11 algorithm based on adaptive downsampling(ADown) module and dynamic convolution(DynamicConv) module to improve breast tumor ultrasound image detection. Methods A YOLOv11-ADown-DynamicConv algorithm was established by the improvment of the original YOLOv11 algorithm with the convolution(Conv) modules in the Backbone and Neck modules replaced by the ADown modules and the Bottleneck module introduced into the DynamicConv module. The YOLOv11-ADown-DynamicConv algorithm was trained and validated with the Breast Ultrasound Image(BUSI) dataset published in Kaggle platform. Results The YOLOv11-ADown-DynamicConv algorithm gained advantages over the original YOLOv11 algorithm when used for breast tumor ultrasound image detection, which had the mean average precision being 0.763 in case of the intersection over union(IoU) threshold of 0.5(mAP50) and 0.522 in case of the intersection over union(IoU) threshold between 0.5 and 0.95(mAP50-95), and the number of floating-point operations and the number of frames per second being 4.8G and 135 frames/s. Conclusion The YOLOv11-ADown-DynamicConv algorithm enhances the YOLOv11 algorithm for breast tumor ultrasound image detection, and facilitates the physician efficiently in disease screening and evaluation. [Chinese Medical Equipment Journal,2025,46(12):1-8]
[1]GENG X,PU Y,HU Z,et al. Effect of intraoperative intravenous infusion of esketamine combined with dexmedetomidine on postoperative sleep disturbance in patients undergoing radical mastectomy[J]. Drug Des Devel Ther,2025,19:4 629-4 640.
[2]GHOSH R,SIL D,SHARMA R,et al. Investigating postbiotics as innovative adjuvants:deciphering the gut-breast connection in breast cancer therapy,from gut microbiome to personalized medicine[J]. Mol Biol Rep,2025,52(1):547.
[3]KARIMI M A,KHADEM R,HARAJ A,et al. The impact of sodium-glucose cotransporter-2 inhibitors on breast cancer and cancer-related mortality:a systematic review and metaanalysis of randomized controlled trials[J]. BMC Cancer,2025,25(1):991.
[4]HSIEH Y H,HSU F R,DAI S T,et al. Incorporating the breast imaging reporting and data system lexicon with a fully convolutional network for malignancy detection on breast ultrasound[J]. Diagnostics(Basel),2021,12(1):66.
[5]GUO Y,CHEN M,YANG L,et al. A neural network with a human learning paradigm for breast fibroadenoma segmentation in sonography[J]. Biomed Eng Online,2024,23(1):5.
[6]FU R,WANG X W,WANG S Y,et al. PMJDM:a multi-task joint detection model for plant disease identification[J]. Front Plant Sci,2025,16:1599671.
[7]罗锋，陈锋，吴航，等.基于Faster R-CNN的损害控制手术器械检测方法研究[J].医疗卫生装备，2023,44(6):1-6.
[8]ZHANG Z,ZHANG X,LIN X,et al. Ultrasonic diagnosis of breast nodules using modified Faster R-CNN[J]. Ultrason Imaging,2019,41(6):353-367.
[9]XIA Z,YUAN R,CAO Y,et al. A systematic review of the application of machine learning techniques to ultrasound tongue imaging analysis[J]. J Acoust Soc Am,2024,156(3):1 796-1 819.
[10]VELLAN C J,NATESAN V,JAYAPALAN J J. Metabolomics biomarkers for breast cancer:an updated perspective[J]. Clin Breast Cancer,2025,25(6):e765-e778.
[11]WANG C,HAN Y,YANG C,et al. CF-YOLO for small target detection in drone imagery based on YOLOv11 algorithm[J]. Sci Rep,2025,15(1):16741.
[12]TENG H,WANG Y,LI W,et al. Advancing rice disease detection in farmland with an enhanced YOLOv11 algorithm[J].Sensors(Basel),2025,25(10):3056.
[13]WANG H,LU J,YANG S,et al. Cervical vertebral body segmentation in X-ray and magnetic resonance imaging based on YOLO-UNet:automatic segmentation approach and available tool[J]. Digit Health,2025,11:20552076251347695.
[14]TILL T,SCHERKL M,STRANGER N,et al. Impact of test set composition on AI performance in pediatric wrist fracture detection in X-rays[J]. Eur Radiol,2025,35(11):6 853-6 864.
[15]CHENG C,CHENG X,LI D,et al. Drill pipe detection and counting based on improved YOLOv11 and Savitzky-Golay[J]. Sci Rep,2025,15(1):16779.
[16]WU H,GUAN M,CHEN J,et al. OE-YOLO:an EfficientNetbased YOLO network for rice panicle detection[J]. Plants(Basel),2025,14(9):1370.
[17]HE L,ZHOU Y,LIU L,et al. Application of the YOLOv11-seg algorithm for AI-based landslide detection and recognition[J]. Sci Rep,2025,15(1):12421.
[18]HE L H,ZHOU Y Z,LIU L,et al. Research on object detection and recognition in remote sensing images based on YOLOv11[J]. Sci Rep,2025,15(1):14032.
[19]贺永强，齐珊，李泽萌，等.基于YOLOv5的眼前节参数实时测量算法研究[J].医疗卫生装备，2023,44(6):7-13.
[20]RAJ A,DAWALE M,WAYAL S,et al. YOLO-ODD:an improved YOLOv8s model for onion foliar disease detection[J].Front Plant Sci,2025,16:1551794.
[21]谢骏，平钦文，曹濒月，等.基于改进YOLOv8的无人机可见光小目标检测方法研究[J].医疗卫生装备，2025,46(1):1-6.
[22]PAN Y,ZHANG Z,ZHANG X,et al. YOLO-TARC:YOLOv10with token attention and residual convolution for small void detection in root canal X-ray images[J]. Sensors(Basel),2025,25(10):3036.
[23]SAZAK H,KOTAN M. Automated blood cell detection and classification in microscopic images using YOLOv11 and optimized weights[J]. Diagnostics(Basel),2024,15(1):22.
[24]BHAGWAT A,DUTTA S,SAHA D,et al. An online 11 kv distribution system insulator defect detection approach with modified YOLOv11 and mobileNetV3[J]. Sci Rep,2025,15(1):15691.
[25]XIE Y,ZHU B,JIANG Y,et al. Diagnosis of pneumonia from chest X-ray images using YOLO deep learning[J]. Front Neurorobot,2025,19:1576438.
[26]YUAN W,YANG C,WANG X,et al. CV-YOLOv10-AR-M:foreign object detection in Pu-erh tea based on five-fold cross-validation[J]. Foods,2025,14(10):1680.
[27]XUE B,ZHANG B,CHENG Q. Experiment study on UAV target detection algorithm based on YOLOv8n-ACW[J]. Sci Rep,2025,15(1):11352.
[28]ZHOU M,WAN X,YANG Y,et al. EBR-YOLO:a lightweight detection method for non-motorized vehicles based on drone aerial images[J]. Sensors(Basel),2025,25(1):196.
基本信息:
DOI：10.19745/j.1003-8868.2025211
中图分类号:TP391.41;R737.9
引用信息:
[1]洪成坤,王晓阳,付丽媛.基于改进YOLOv11算法的超声图像乳腺肿瘤检测研究[J].医疗卫生装备,2025,46(12):1-8.DOI:10.19745/j.1003-8868.2025211.
基金信息:
福建省科技计划项目（2021I0037）