基于改进YOLOv7的肝囊型包虫病超声图像小病灶检测
Small lesion detection in ultrasound images of hepatic cystic echinococcosis based on improved YOLOv7
米吾尔依提·海拉提 1热娜古丽·艾合麦提尼亚孜 1卡迪力亚·库尔班 1严传波2
作者信息
- 1. 新疆医科大学公共卫生学院,新疆乌鲁木齐 830011
- 2. 新疆医科大学医学工程技术学院,新疆乌鲁木齐 830011
- 折叠
摘要
目的:提出一种基于YOLOv7用于检测肝囊型包虫病超声图像小病灶的方法.方法:首先用轻量级特征提取主干网络GhostNet替换原特征提取主干,降低模型总参数量;其次为改善YOLOv7网络的评价指标CIoU在作为损失函数时,检测精度较低的问题,用更优的ECIoU替换CIoU,进一步提高模型检测精度.结果:在自建的肝囊型包虫病超声图像小病灶数据集上进行训练,结果显示改进后的模型大小为59.4 G,mAP@0.5检测精度为88.1%,相比原始的模型性能得到提升,并超过其余主流检测方法.结论:本文模型能更高效地检测并分类肝囊型包虫病超声图像中的病灶位置和类别.
Abstract
Objective To propose a novel algorithm model based on YOLOv7 for detecting small lesions in ultrasound images of hepatic cystic echinococcosis.Methods The original feature extraction backbone was replaced with a lightweight feature extraction backbone network GhostNet for reducing the quantity of model parameters.To address the problem of low detection accuracy when the evaluation index CIoU of YOLOv7 was used as a loss function,ECIoU was substituting for CIoU,which further improved the model detection accuracy.Results The model was trained on a self-built dataset of small lesion ultrasound images of hepatic cystic echinococcosis.The results showed that the improved model had a size of 59.4 G and a detection accuracy of 88.1%for mAP@0.5,outperforming the original model and surpassing other mainstream detection methods.Conclusion The proposed model can detect and classify the location and category of lesions in ultrasound images of hepatic cystic echinococcosis more efficiently.
关键词
囊型包虫病/深度学习/目标检测/YOLOv7/ECIoU/GhostNetKey words
cystic echinococcosis/deep learning/object detection/YOLOv7/ECIoU/GhostNet引用本文复制引用
基金项目
国家自然科学基金(81560294)
省部共建中亚高发病成因与防治国家重点实验室(SKL-HIDCA-2020-YG)
出版年
2024
NETL
NSTL
万方数据