WIP 64 - Calculation Of Lung volume In Neonates (COLIN Study)

Poster Number: WIP 64
Publication Number: WIP 64.716

Background: The number of ribs visible above the diaphragm on chest X-rays (CXR, diaphragm position) is frequently used by clinicians to assess lung volume in neonates requiring respiratory support. No large study has validated this method against any gold-standard calculation of lung volume, raising the possibility that this practice is misinforming clinical decisions.

Objective: To describe the relationship between diaphragm position on Computed Tomography (CT) topogram (scout CXR) against total aerated lung volume (VL; ml/kg) calculated from the same CT in neonates.

Design/Methods: This study is approved by the Institution Review Board of The Royal Children’s Hospital (RCH), Melbourne, Australia (RCH HREC 97964), and prospectively registered (ANZCTR 12523000903684).

CT chest scans performed within the first 31 days of life (2015-2023) were identified in the Medical Imaging database at RCH. Scans with congenital lung pathology, single ventricle anatomy, significant lordosis or scoliosis or a cardiothoracic ratio >0.6 were excluded. A total of 259 CT scans were analysed.

VL (right, left and total) was calculated from CT using the syngo.via (Siemens Healthcare GmbH, Germany) semi-automated tissue segmentation tool. Assessment of the first 30 scans showed excellent agreement; Intraclass Correlation Coefficients 0.86-0.99. Diaphragm position and distance (lung apex to diaphragm; mm) on de-identified topograms was assessed by three neonatologists.

The distribution of VL (continuous) and diaphragm position (ordinal categorical) will be presented with descriptive statistics (Box and Whisker plot of VL at each rib level) with further correlation and linear regression analysis if appropriate. Secondary outcomes include reference to Hounsfield units (measure of density) and subgroup analysis by gestation, CT indication and mode of respiratory support. Data analysis will be performed in the R statistical software package.

Data entry has been completed for 115 of the 259 scans, with expected completion by December 2023 and data analysis finalised by February 2024.