WS07.02Investigating imaging biomarkers in cystic fibrosis preschoolers using manual and Artificial intelligence-based algorithms

IF 5.4 2区医学 Q1 RESPIRATORY SYSTEM

Journal of Cystic Fibrosis Pub Date : 2025-06-01 DOI:10.1016/j.jcf.2025.03.529

P. Raut , M. Bonte , B. Manai , P. Makani , D. Caudri , H.M. Janssens

{"title":"WS07.02Investigating imaging biomarkers in cystic fibrosis preschoolers using manual and Artificial intelligence-based algorithms","authors":"P. Raut , M. Bonte , B. Manai , P. Makani , D. Caudri , H.M. Janssens","doi":"10.1016/j.jcf.2025.03.529","DOIUrl":null,"url":null,"abstract":"<div><h3>Introduction</h3><div>Monitoring structural lung changes in preschoolers with cystic fibrosis (PrewCF) is challenging, but CT could be used to quantify structural lung changes. This study evaluates and compares manual vs automated image analysis tools as possible sensitive biomarkers in PrewCF.</div></div><div><h3>Methods</h3><div>76 Erasmus MC-Sophia PrewCF (1-5 yrs) were included, with 177 CT's. Manual PRAGMA-CF score include %Bronchiectasis (%BE), %Mucus plugging (%MP), %Airway wall thickening (%AWT), and overall %Disease (%DIS). With the AI-based LungQ<sup>TM</sup> platform (Thirona) automated PRAGMA-AI was used to calculate %BE, %AWT and %DIS. Bronchial dimensions were assessed with the Bronchus-Artery (BA) method, and number of mucus plugs with the Mucus plugging (MP) algorithm. Manual and automated outcomes were compared using correlation, ICC and multivariable regression. Finally, progression over time was evaluated in 41 PrewCF with >1 CT scans using both methods.</div></div><div><h3>Results</h3><div>All 177 CT's were scored manually, 126 (71%) also processed successfully with LungQ (free-breathing: 64%; inspiratory: 98%). AWT was the most prevalent abnormality. Correlations between manual and automatic PRAGMA analysis were strong for %DIS (0.67, CI:0.56-0.75, p<0.001), moderate for %BE (0.53, CI:0.39-0.65, p<0.001) and %AWT (0.59, CI:0.46-0.69, p<0.001). ICC's showed good agreement for %DIS (0.77, CI:0.69-0.83), moderate for %AWT (0.58, CI:0.42-0.70) and weak for %BE (0.35, CI:0.18-0.50). In a backward stepwise multivariable model automated outer bronchial diameter (Bout/A ratio; β=2.54, p=0.01) and mucus plugs (β=0.18, p<0.001) remained independent significant predictors of manual %DIS. Significant progression over time was detected using the manual %DIS and %AWT, as well as automated %DIS and bronchial wall thickness (Bwt/A ratio).</div></div><div><h3>Conclusion</h3><div>Both manual and automated CT analysis are feasible in PrewCF and methods show reasonable agreement. Both methods were able to detect significant disease progression over a 2-4 year period</div></div>","PeriodicalId":15452,"journal":{"name":"Journal of Cystic Fibrosis","volume":"24 ","pages":"Page S14"},"PeriodicalIF":5.4000,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Cystic Fibrosis","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1569199325006253","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RESPIRATORY SYSTEM","Score":null,"Total":0}

引用次数: 0

Abstract

Introduction

Monitoring structural lung changes in preschoolers with cystic fibrosis (PrewCF) is challenging, but CT could be used to quantify structural lung changes. This study evaluates and compares manual vs automated image analysis tools as possible sensitive biomarkers in PrewCF.

Methods

76 Erasmus MC-Sophia PrewCF (1-5 yrs) were included, with 177 CT's. Manual PRAGMA-CF score include %Bronchiectasis (%BE), %Mucus plugging (%MP), %Airway wall thickening (%AWT), and overall %Disease (%DIS). With the AI-based LungQ^TM platform (Thirona) automated PRAGMA-AI was used to calculate %BE, %AWT and %DIS. Bronchial dimensions were assessed with the Bronchus-Artery (BA) method, and number of mucus plugs with the Mucus plugging (MP) algorithm. Manual and automated outcomes were compared using correlation, ICC and multivariable regression. Finally, progression over time was evaluated in 41 PrewCF with >1 CT scans using both methods.

Results

All 177 CT's were scored manually, 126 (71%) also processed successfully with LungQ (free-breathing: 64%; inspiratory: 98%). AWT was the most prevalent abnormality. Correlations between manual and automatic PRAGMA analysis were strong for %DIS (0.67, CI:0.56-0.75, p<0.001), moderate for %BE (0.53, CI:0.39-0.65, p<0.001) and %AWT (0.59, CI:0.46-0.69, p<0.001). ICC's showed good agreement for %DIS (0.77, CI:0.69-0.83), moderate for %AWT (0.58, CI:0.42-0.70) and weak for %BE (0.35, CI:0.18-0.50). In a backward stepwise multivariable model automated outer bronchial diameter (Bout/A ratio; β=2.54, p=0.01) and mucus plugs (β=0.18, p<0.001) remained independent significant predictors of manual %DIS. Significant progression over time was detected using the manual %DIS and %AWT, as well as automated %DIS and bronchial wall thickness (Bwt/A ratio).

Conclusion

Both manual and automated CT analysis are feasible in PrewCF and methods show reasonable agreement. Both methods were able to detect significant disease progression over a 2-4 year period

查看原文本刊更多论文

ws07.02使用人工和人工智能算法研究囊性纤维化学龄前儿童的成像生物标志物

监测患有囊性纤维化（PrewCF）的学龄前儿童肺结构性变化具有挑战性，但CT可用于量化肺结构性变化。本研究评估和比较了手动和自动图像分析工具作为PrewCF可能敏感的生物标志物。方法选取年龄1 ~ 5岁的Erasmus MC-Sophia PrewCF患者76例，CT扫描177张。手动PRAGMA-CF评分包括%支气管扩张（%BE）、%粘液堵塞（%MP）、%气道壁增厚（%AWT）和总体%疾病（%DIS）。采用基于人工智能的lunqtm平台（Thirona）自动化PRAGMA-AI计算%BE、%AWT和%DIS。支气管动脉（BA）法评估支气管尺寸，黏液堵塞（MP）法评估黏液堵塞数量。使用相关性、ICC和多变量回归对人工和自动结果进行比较。最后，通过两种方法的1次CT扫描评估41例PrewCF随时间的进展情况。结果177个CT全手工评分，126个（71%）用LungQ评分成功(自由呼吸：64%；吸入:98%)。AWT是最常见的异常。手动和自动PRAGMA分析之间的相关性在%DIS （0.67, CI:0.56-0.75, p<0.001）中很强，在%BE （0.53, CI:0.39-0.65, p<0.001）和%AWT （0.59, CI:0.46-0.69, p<0.001）中中等。ICC对%DIS （0.77, CI:0.69-0.83）、%AWT （0.58, CI:0.42-0.70）和%BE （0.35, CI:0.18-0.50）的一致性较好。在后向逐步多变量模型中，自动支气管外直径(Bout/ a ratio；β=2.54, p=0.01)和粘液塞（β=0.18, p= 0.001）仍然是手工%DIS的独立显著预测因子。使用手动%DIS和%AWT，以及自动%DIS和支气管壁厚度（Bwt/A比）检测到随时间的显著进展。结论人工和自动CT分析在PrewCF中均是可行的，且方法一致。两种方法都能够在2-4年的时间内检测到显著的疾病进展

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Journal of Cystic Fibrosis 医学-呼吸系统

CiteScore

10.10

自引率

13.50%

发文量

1361

审稿时长

50 days

期刊介绍： The Journal of Cystic Fibrosis is the official journal of the European Cystic Fibrosis Society. The journal is devoted to promoting the research and treatment of cystic fibrosis. To this end the journal publishes original scientific articles, editorials, case reports, short communications and other information relevant to cystic fibrosis. The journal also publishes news and articles concerning the activities and policies of the ECFS as well as those of other societies related the ECFS.