Optimizing Photon Counting CT imaging for Interstitial Lung Disease assessment

Résumé fourni par la source

AIMS: To optimize photon-counting detector CT(PCD-CT) imaging protocols for interstitial lung disease(ILD) by evaluating the impact of reconstruction parameters on both visual and automatic assessment. METHODS: PCD-CT scans from ILD patients were reconstructed using n=72 different combinations of CT imaging parameters by modifying the following reconstruction parameters: kernel type (Bl, Br, Qr), kernel resolution (56, 60, 64), slice thickness (0.2, 0.6, 1mm) and iterative reconstruction levels (Q1, Q2, Q3). Six thoracic radiologists and five pulmonologists visually assessed ILD abnormalities to determine the best CT parameters to evaluate ILD features. LungQTM software (Thirona, Nijmegen) performed automatic quantification of ILD abnormalities (%ILD). Phantom scans evaluated noise, in-plane resolution and contrast-to-noise ratio (CNR) comparing PCD-CT and energy-integrating detector (EID-CT) scanners. RESULTS: Bl64 kernel with Q3 and 0.6mm slice thickness was the preferred setting for visual assessment. Automatic analysis revealed the greatest variation of %ILD with Bl64 at 0.2mm (SD=22.39, range: 2.14–69.68%). Smoother kernels and thicker slices reduced %ILD variability.. Phantom scans showed that despite thinner slice and fixed dose and kernels, PCD-CT had lower noise (SD 99.29 vs. 102.04), superior resolution (1.104 1/mm vs. 0.821 1/mm), and higher CNR (8.91 vs. 8.44) than EID-CT. CONCLUSIONS: Optimal setting for visualizing ILD abnormalities was Bl64 kernel with 0.6 mm slice thickness. Smaller slice thicknesses and sharper kernels result in significant variability in automatic ILD quantification that must be taken in account for both visual and automatic assessments, especially during ILD follow-up.