Inter-institutional variability in CT-to-mass-density conversion tables for helical tomotherapy: a national survey in Japan

TL;DR

This study examines differences in CT-to-mass-density conversion practices across Japanese institutions using helical tomotherapy and suggests ways to improve standardization and quality.

Contribution

The study identifies inter-institutional variability in CT–MD tables and proposes standardization strategies for adaptive radiotherapy.

Findings

SimCT tables showed the most variation in data points and high-density CT numbers.

Half-phantom setups reduced variability in ClearRT tables.

CTTrue tables showed high consistency but had inconsistent air CT number handling.

Abstract

This study evaluates current practices and challenges associated with computed tomography number-to-mass density (CT–MD) conversion tables in helical tomotherapy across Japan and explores directions for standardization and quality improvement amid the increasing adoption of adaptive radiotherapy (ART). A nationwide web-based survey was conducted across 34 institutions utilizing the Radixact system. Data were collected on CT acquisition protocols, calibration phantoms, density plugs, reconstruction algorithms, table registration timing and quality assurance (QA) frequency. Registered CT–MD tables were categorized by CT modality: Simulation CT (SimCT), ClearRT and CTrue. ClearRT tables were analyzed by phantom setup (full vs half), and CTrue tables by reconstruction method [filtered back projection (FPB) vs iterative reconstruction (IR)]. Inter-institutional variations in CT numbers and the number of data points were assessed. SimCT tables exhibited the widest variation in the number of data points (median = 10) and high-density CT numbers. ClearRT tables (median = 8) showed variations of up to 300 Hounsfield units (HU) in cortical bone; the half-phantom setup reduced inter-institutional variability. CTrue tables (median = 8) demonstrated high consistency, with negligible differences between IR and FPB. All plug CT numbers of CTrue remained within the tolerance defined by the American Association of Physicists in Medicine Task Group 148. However, CT numbers for air plugs varied by ~±30 HU, indicating inconsistent handling of air reference values. Additionally, 43% of institutions did not perform routine QA. Standardizing phantom geometry, air CT number handling and QA protocols—particularly using half-phantom calibration—may improve CT–MD table consistency and dose accuracy in ART.

Figures