Impact of choice of volumetry software and nodule management guidelines on recall rates in lung cancer screening. Issue 120 (November 2019)

• Record Type: Journal Article
• Authors: Soo, E.; Edey, A.J.; Mak, S.M.; Moser, J.; Mohammadi, S.; Rodrigues, T.; Duffy, S.W.; Field, J.K.; Baldwin, D.R.; Nair, A.; Devaraj, A.
• Is Part Of: European journal of radiology. Issue 120(2019)
• Publication Date: 2019-11
• Subjects: CT computed tomography -- VSP volume software package -- BTS British Thoracic Society -- UKLS United Kingdom Lung Cancer screening; Lung cancer screening -- Pulmonary nodule -- Volumetry -- Computed tomography; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2019.108646
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: Volumetry is the preferred method to assess nodule size in lung cancer screening. Volumetry application suitability for screening depends on the chosen guideline. Nodule size overestimation by one software package led to excessive recall rates. Abstract: Purpose: Appropriate lung nodule management is essential to minimizing unnecessary patient recall in lung cancer screening. Two European guidelines provide differing recommendations in that participants with nodules ≥100 mm 3 or ≥80 mm 3 respectively should be recalled, at baseline. Nodule size estimation is known to vary between volumetry software packages (VSPs). The aim of this study was to examine the impact of choice of VSP on participant recall rates, when applying different European nodule management guidelines. An additional aim was to compare recall rates between 7 VSPs and manual diameter measurements. Methods: 156 small-sized lung nodules (50–150 mm 3 ) from the UK Lung Screening trial were measured using 7 different VSPs (VSP1-7) and also using manual diameter. The type of VSP used in the NELSON study (VSP1), on which European nodule management guidelines are based, provided the reference standard. Nodule size was compared using Bland Altman, and recall rates by Mcnemar's test. Results: Compared to the reference standard, a 100 mm 3 threshold for recall, resulted in no difference in recall rates only for VSP 5 & 7. Using an 80mm 3 threshold resulted in no difference in recall rates for VSP2 & 6. Recall rates were significantly higher for VSP 4 regardless of threshold and when using manual diameter measurements. Conclusions: Appropriate nodule size thresholds for recall in screening depend on the type of volumetry software used. The results highlight the importance of benchmarking of volumetry packages.