基于AHP-FCE的胸部后前位图像质量控制评价体系的建立及应用

王振邦; 罗维; 郑君惠; 区杰浩; 曹希明; 钟小梅

doi:10.12019/j.issn.1671-5144.2023.01.007

基于AHP-FCE的胸部后前位图像质量控制评价体系的建立及应用

Establishment and Application of Posteroanterior Chest Radiograph Quality Control Evaluation System Based on AHP-FCE

摘要

摘要:
目的探讨一种在评价要素上具有不同权重及将其量化的胸部后前位图像质量控制评价体系的可行性。
方法收集胸部后前位图像评价要素的相关资料，通过层次分析法（analytic hierarchy process，AHP）和模糊综合评价（fuzzy comprehensive evaluation，FCE）建立胸部后前位图像AHP-FCE评价体系，应用该评价体系对从不同设备上抽取的60份胸部后前位X线摄影图像进行评价，最后对评价结果进行统计分析以明确该评价体系的效能。
结果评价体系划分为三个模块，分别是：体位评价、图像后处理评价、图像总体评价，其权重指数依次为：0.419 7、0.311 0、0.269 4。权重的一致性通过一致性比率（consistent ratio，CR）值检测，各要素的CR值介于0.000 0~0.098 2之间，CR值均小于0.10，反映矩阵的合理性。评价结果（评分）的一致性检验包括：各专家组内相关系数（interclass correlation coefficient，ICC）=0.882，P < 0.001、同一专家重复评分（专家A：ICC=0.867、专家B：ICC=0.915，P < 0.001），各年龄段：20~40岁（ICC=0.895，P < 0.001）、41~60岁（ICC=0.859，P < 0.001）、61~80岁（ICC=0.890，P < 0.001），性别：男性（ICC=0.829，P < 0.001）、女性（ICC=0.924，P < 0.001），阴/阳性：阴性（ICC=0.826，P < 0.001）、阳性（ICC=0.893，P < 0.001），机型：GE definium 6000机型的评分（ICC=0.887，P < 0.001）、Philips Digital Diagnost TH+VS机型的评分（ICC=0.919，P < 0.001）、Siemens Aristos Axiom机型的评分（ICC=0.837，P < 0.001），都能反映较好的一致性。个人与总体评价级别的一致性检验中等（Kappa=0.41，P < 0.001），说明一致性一般。
结论本研究通过AHP-FCE建立的新胸部后前位评价体系，可以反映不同评价因素对图像质量的重要性并将评价量化，相对能够更客观地反映图像的质量，评价结果一致性较好、可重复性高，具有较高的实用性。

Abstract:
Objective To explore the feasibility of a posteroanterior (PA) chest radiograph quality control evaluation system with different weights and quantification in the evaluation elements.
Methods The related data of PA chest radiograph evaluation elements were collected, and the PA chest radiograph analytic hierarchy process (AHP)-fuzzy comprehensive evaluation (FCE) evaluation system was established by AHP and FCE. The system evaluated 60 PA chest radiographs extracted from different devices and finally conducted a statistical analysis of the evaluation results to clarify the effectiveness of the evaluation system.
Results The evaluation system was divided into three modules, which were body position evaluation, image post-processing evaluation, and image overall evaluation. The weight indexes were as follows: 0.419 7, 0.311 0, and 0.269 4, respectively. The consistency of the weights was tested by consistent ratio (CR) value. The CR value of each element was between 0.000 0~0.098 2, and the CR value was less than 0.10, reflecting the rationality of the matrix. The consistency test of the evaluation results (scoring) included each expert (interclass correlation coefficient (ICC)=0.882, P < 0.001; repeated scoring by the same expert (expert A: ICC=0.867, expert B: ICC=0.915, P < 0.001); each age group: 20~40 years old (ICC=0.895, P < 0.001), 41~60 years old (ICC=0.859, P < 0.001), 61~80 years old (ICC=0.890, P < 0.001); gender: male (ICC=0.829, P < 0.001), female (ICC=0.924, P < 0.001); negative/positive: negative (ICC=0.826, P < 0.001), positive (ICC=0.893, P < 0.001); model: GE definium 6000 Type score (ICC=0.887, P < 0.001), Philips Digital Diagnost TH+VS model score (ICC=0.919, P < 0.001), Siemens Aristos Axiom model score (ICC=0.837, P < 0.001); all can reflect good consistency. The consistency test of individual and overall evaluation levels showed average consistency (Kappa=0.41, P < 0.001).
Conclusions The new PA chest radiograph evaluation system established by AHP-FCE in this study can reflect the importance of different evaluation factors on image quality and quantify the evaluation. It can reflect the image quality more objectively, and the evaluation results were consistent and reliable, with high repeatability and high practicability.

HTML全文

参考文献(20)

施引文献

资源附件(0)