中国中西医结合影像学杂志2024，Vol.22Issue(4)：406-412,7.DOI:10.3969/j.issn.1672-0512.2024.04.008
基于深度学习的X线腰椎正侧位片的智能质量控制研究
Research on intelligent quality control of lumbar spine radiograph based on deep learning
摘要
Abstract
Objective:To explore the feasibility of intelligent quality control method of lumbar spine X-rays based on deep learning technology in clinical work.Methods:A retrospective analysis of 4,690 cases of X-ray lumbar spine anteroposterior and lateral images was conducted.The image quality characteristics were classified into two categories,the image technical quality and image clarity quality,and a quality control knowledge graph was constructed.Through data annotation and model training,the differences between intelligent quality control and manual quality control were evaluated using mean average precision(mAP)and mean absolute error(MAE).Results:According to the reference standard,in terms of the image technical quality,the mAP of intelligent quality control method was significantly better than that of single-person quality control(A,B)and closed to that of multi-person quality control method.In terms of image clarity,the MAE of intelligent quality control method was significantly better than that of single-person and multi-person quality control methods.Conclusions:The intelligent quality control method for lumbar spine X-rays based on deep learning is superior to single-person quality control method,and also superior in image clarity quality to multi-person quality control method.This method can objectively evaluate image quality,improve work efficiency,and will be the potential for widespread application in clinical quality control.关键词
深度学习/知识图谱/腰椎正侧位片/质量质控/智能质量控制Key words
Deep learning/Knowledge graph/Lumbar spine radiograph/Quality control/Intelligent quality control引用本文复制引用
张禹萱,谷宗运,鲁文豪,王倩,宋亮亮,李传富..基于深度学习的X线腰椎正侧位片的智能质量控制研究[J].中国中西医结合影像学杂志,2024,22(4):406-412,7.基金项目
安徽省高校协同创新项目(GXXT-2021-065,GXXT-2022-031). （GXXT-2021-065,GXXT-2022-031）
