| 212 | 0 | 155 |
| 下载次数 | 被引频次 | 阅读次数 |
针对原始YOLOv8算法甲状腺结节检测精度低的问题，提出一种改进的YOLOv8甲状腺结节检测算法。该算法通过在原始算法的基础上添加高效通道注意力（efficient channel attention,ECA）和选择性注意力（large selective kernel,LSK）模块，从不同的维度学习图像特征，更好地捕获甲状腺结节信息。引入空间金字塔多尺度特征融合（spatial pyramid pooling enhanced with elan,SPPELAN）模块并替换backbone中的快速空间金字塔池化（spatial pyramid pooling fast,SPPF），实现甲状腺结节中局部特征和全局特征的融合，准确地定位甲状腺结节。将双向特征金字塔网络（concatenated bidirectional feature pyramid network,Concat-BiFPN）应用于head层，使语义信息传递到不同的特征尺度上，以适应不同形状和尺寸的结节。将损失函数替换为Focal损失函数，有效解决了甲状腺结节检测中样本不平衡的问题。实验结果表明：改进的YOLOv8算法在甲状腺结节检测中效果显著，精度均值提高了1.5%，召回率提高了2%，模型的参数量减少了1.3%。
Abstract：Aiming at the low detection accuracy of the original YOLOv8 algorithm for thyroid nodules, this paper proposes an improved YOLOv8 algorithm for thyroid nodule detection. By adding efficient channel attention(ECA) and large selective kernel(LSK) modules to the original algorithm, image features are learnt from various dimensions to better capture the information of thyroid nodules. The spatial pyramid pooling enhanced with Elan(SPPELAN) module is introduced, replacing spatial pyramid pooling fast(SPPF) in the backbone to achieve the fusion of local and global features in thyroid nodules for accurate localization of thyroid nodules. The concatenated bidirectional feature pyramid network(concat-BiFPN) is applied to the head layer so that the semantic information is delivered to different feature scales for nodules of different shapes and sizes. Finally, the loss function is replaced with the Focal loss function, which effectively handles the problem of sample imbalance in thyroid nodule detection. Experimental results show that the improved YOLOv8 algorithm obtains significant results in thyroid nodule detection, with a 1.5% improvement in accuracy average, a 2% improvement in recall, and a 1.3%reduction in the number of the model′s parameters.
[1]YANG J,SHI X,WANG B,et al.Ultrasound image classification of thyroid nodules based on deep learning[J].Front Oncol,2022,12:905955.
[2]HAUGEN B R,ALEXANDER E K,BIBLE K C,et al.2015American thyroid association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer:the American thyroid association guidelines task force on thyroid nodules and differentiated thyroid cancer[J].Thyroid,2016,26(1):1-133.
[3]郑天雷，杨娜，耿诗，等.一种基于Faster R-CNN的甲状腺结节超声图像目标检测改进算法[J].四川大学学报：医学版，2023,54(5):915-922.
[4]LIN X,ZHOU X,TONG T,et al.A super-resolution guided network for improving automated thyroid nodule segmentation[J].Comput Methods Programs Biomed,2022,227:107186.
[5]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Trans Pattern Anal Mach Intell,2017,39(6):1137-1149.
[6]LIU W,ANGUELOV D,ERHAN D,et al.SSD:Single shot multibox detector[C]//European Conference on Computer Vision.Cham:Springer,2016:21-37.
[7]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//2016IEEE Conference on Computer Vision and Pattern Recognition(CVPR).USA:IEEE,2016:779-788.
[8]柯威.基于深度卷积神经网络的甲状腺乳头状癌超声图像识别的研究[D].广州：广东工业大学，2018.
[9]SONG W,LI S,LIU J,et al.Multitask cascade convolution neural networks for automatic thyroid nodule detection and recognition[J].IEEE J Biomed Health Inform,2019,23(3):1215-1224.
[10]WANG L,YANG S,YANG S,et al.Automatic thyroid nodule recognition and diagnosis in ultrasound imaging with the YOLOv2 neural network[J].World J Surg Oncol,2019,17(1):12.
[11]ZHANG L,ZHUANG Y,HUA Z,et al.Automated location of thyroid nodules in ultrasound images with improved YOLOV3 network[J].J Xray Sci Technol,2021,29(1):75-90.
[12]YANG D Q,XIAJ F,LI R Z,et al.Automatic thyroid nodule detection in ultrasound imaging with improved YOLOv5 neural network[J].IEEE Access,2024,12:22662-22670.
[13]LI X X,YIN P P,QU Y Q,et al.Research on multi-object tracking algorithm for thyroid nodules based on Byte Track[C]//2024 IEEE 3rd International Conference on Electrical Engineering,Big Data and Algorithms (EEBDA).Changchun:IEEE,2024:1589-1592.
[14]WANG H,SONG Z L.Improved mosaic:algorithms for more complex images[J].Journal of Physics:Conference Series,2020,1684(1):012094.
[15]HE Q,XU A,YE Z,et al.Object detection based on lightweight YOLOX for autonomous driving[J].Sensors (Basel),2023,23(17):7596.
[16]GHIASI G,LIN T Y,LE Q V.NAS-FPN:learning scalable feature pyramid architecture for object detection[C]//2019IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).USA:IEEE,2019:7036-7045.
[17]YANG J F,FU X Y,HU Y W,et al.Pan Net:a deep network architecture for pan-sharpening[C]//2017 IEEE International Conference on Computer Vision(ICCV).Venice:IEEE,2017:5449-5457.
[18]LI Y X,HOU Q B,ZHENG Z H,et al.Large selective kernel network for remote sensing object detection[C]//2023 IEEE/CVF International Conference on Computer Vision(ICCV).France:IEEE,2023:16794-16805.
[19]WANG Q L,WU B G,ZHU P F,et al.ECA-net:efficient channel attention for deep convolutional neural networks[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).USA:EEE,2020:11534-11542.
[20]TAN M X,PANG R M,LE Q V.Efficient Det:scalable and efficient object detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognitio (CVPR).USA:IEEE,2020:10781-10790.
[21]LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detectio[J].IEEE Trans Pattern Anal Mach Intell,2020,42(2):318-327.
基本信息:
DOI：10.19573/j.issn2095-0926.202404007
中图分类号:R581;TP183;TP391.41
引用信息:
[1]王利莎,刘芬.改进的YOLOv8甲状腺结节目标检测算法[J].天津职业技术师范大学学报,2024,34(04):38-45.DOI:10.19573/j.issn2095-0926.202404007.
基金信息:
天津市科技特派员项目（22YDTPJC00330）
2024-12-28
2024-12-28