Automated assessment of functional lung imaging with 68Ga-ventilation/perfusion PET/CT using iterative histogram analysis
Publication Name
EJNMMI Physics
Abstract
Purpose: Functional lung mapping from Ga -ventilation/perfusion (V/Q) PET/CT, which has been shown to correlate with pulmonary function tests (PFTs), may be beneficial in a number of clinical applications where sparing regions of high lung function is of interest. Regions of clumping in the proximal airways in patients with airways disease can result in areas of focal intense activity and artefact in ventilation imaging. These artefacts may even shine through to subsequent perfusion images and create a challenge for quantitative analysis of PET imaging. We aimed to develop an automated algorithm that interprets the uptake histogram of PET images to calculate a peak uptake value more representative of the global lung volume. Methods: Sixty-six patients recruited from a prospective clinical trial underwent both V/Q PET/CT imaging and PFT analysis before treatment. PET images were normalised using an iterative histogram analysis technique to account for tracer hotspots prior to the threshold-based delineation of varying values. Pearson’s correlation between fractional lung function and PFT score was calculated for ventilation, perfusion, and matched imaging volumes at varying threshold values. Results: For all functional imaging thresholds, only FEV1/FVC PFT yielded reasonable correlations to image-based functional volume. For ventilation, a range of 10–30% of adapted peak uptake value provided a reasonable threshold to define a volume that correlated with FEV1/FVC (r = 0.54–0.61). For perfusion imaging, a similar correlation was observed (r = 0.51–0.56) in the range of 20–60% adapted peak threshold. Matched volumes were closely linked to ventilation with a threshold range of 15–35% yielding a similar correlation (r = 0.55–0.58). Conclusions: Histogram normalisation may be implemented to determine the presence of tracer clumping hotspots in Ga-68 V/Q PET imaging allowing for automated delineation of functional lung and standardisation of functional volume reporting. 68
Open Access Status
This publication may be available as open access
Volume
8
Issue
1
Article Number
23
Funding Number
APP1038399
Funding Sponsor
National Health and Medical Research Council