计算机科学与探索

• 学术研究 •    下一篇

改进YOLOv8模型实现多类型肺结节检测

包强强,唐思源,李擎乾,王乃钰,杨敏,谷宇,赵金亮,高婧博,王嘉欣,曲禹涵   

  1. 1. 内蒙古科技大学 数智产业学院, 内蒙古 包头 014010
    2. 内蒙古科技大学 包头医学院 计算机科学与技术学院, 内蒙古 包头 014040
    3. 内蒙古科技大学 包头医学院 图书馆, 内蒙古 包头 014040
    4. 内蒙古科技大学 数智产业学院 内蒙古自治区模式识别与智能图像处理重点实验室, 内蒙古 包头 014010

Improve the YOLOv8 model for multi-type lung nodule detection

BAO Qiangqiang,  TANG Siyuan, LI Qingqian, WANG Naiyu, YANG Min, GU Yu, ZHAO Jinliang, GAO Jingbo, WANG Jiaxin, QU Yuhan   

  1. 1. School of Digtial and Intelligence Industry, Inner Mongolia University of Science and Technology, Baotou, Inner Mongolia 014010, China
    2.School of Computer Science and Technology, Inner Mongolia University of Science and Technology Baotou Medical College, Baotou, Inner Mongolia 014040, China
    3. Library, Inner Mongolia University of Science and Technology Baotou Medical College, Baotou, Inner Mongolia 014040, China
    4.Inner Mongolia Key Laboratory of Pattern Recognition and Intelligent Image Processing, School of Digtial and Intelligence Industry, Inner Mongolia University of Science and Technology, Baotou, Inner Mongolia 014010, China

摘要: 目前,肺结节检测通常是对实性肺结节的单一型检测,不同类型肺结节对应多种肺癌类型,多类型检测有助于提高肺癌的整体检出率,提升治愈率。为实现对实性、混合型、磨玻璃型多类型肺结节检测,对YOLOv8模型进行针对性改进。首先,提出RepViTCAA模块对主干部分的C2f模块进行改进,提升微小肺结节检测精度并对模型进行轻量化设计。其次,提出ECLA-HSFPN模块,重建模型特征融合部分,提升尺度不固定肺结节检测精度。接着,将KAN网络融入模型当中,基于KAN网络非线性特征学习能力强的特性,进一步提升微小肺结节检测精度,增强模型泛化能力。最后,基于Inner-IoU辅助框思想,对CIoU损失函数进行改进,进一步解决肺结节尺度不固定问题,提升模型检测精度。在LUNA16数据集中进行测试,改进模型相比原模型及YOLOv9、RT-DETR等主流模型各项评价指标均有提升。在4种类型(实性、磨玻璃型、混合型、微小型)肺结节的专项数据集中测试,改进模型检测效果优于原模型。在LUNA16与本地医院的混合数据集中进行泛化性测试,改进模型具有较强的泛化能力。针对多类型肺结节检测任务,模型的改进较为有效,可以准确检测不同类型的肺结节。

关键词: 多类型肺结节检测, YOLOv8, RepViTCAA, ECLA-HSFPN, KAN, Inner-IoU

Abstract: Currently, lung nodule testing is usually a single type of testing for solid lung nodules, and different kinds of lung nodules correspond to multiple types of lung cancer. Multi-type testing can help improve the overall detection rate of lung cancer and enhance the cure rate. Targeted improvements to the YOLOv8 model were made to enable the detection of multiple types of lung nodules, including solid, mixed, and ground glass. First, the RepViTCAA module is proposed to improve the C2f module of the main part to enhance the accuracy of tiny lung nodule detection and the lightweight design of the model. Second, the ECLA-HSFPN module is proposed to reconstruct the feature fusion part of the model to improve the scale-invariant lung nodule detection accuracy. Then, the KAN network is integrated into the model further to improve the detection accuracy of tiny lung nodules and enhance the generalization ability of the model based on the strong learning ability of nonlinear features of the KAN network. Finally, based on the Inner-IoU auxiliary frame idea, the CIoU loss function is improved to solve the problem of unfixed lung nodule scale and enhance the model detection accuracy. Tested in the LUNA16 dataset, the improved model has improved all evaluation indexes compared with the original model and mainstream models such as YOLOv9 and RT-DETR. The improved model detected better than the original model when tested in a specialized dataset of four types (solid, ground glass, mixed, and microminiature) of lung nodules. The generalizability was tested on a mixed dataset of LUNA16 and local hospitals, and the improved model has strong generalization ability. The improvement of the model is more effective for the task of detecting multiple types of lung nodules and can accurately detect different types of lung nodules.

Key words: Multi-type lung nodule detection, YOLOv8, RepViTCAA, ECLA-HSFPN, KAN, Inner-IoU