Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients.

BACKGROUND: In the Danish Head and Neck Cancer Group (DAHANCA) 35 trial, patients are selected for proton treatment based on simulated reductions of Normal Tissue Complication Probability (NTCP) for proton compared to photon treatment at the referring departments. After inclusion in the trial, immobilization, scanning, contouring and planning are repeated at the national proton centre. The new contours could result in reduced expected NTCP gain of the proton plan, resulting in a loss of validity in the selection process. The present study evaluates if contour consistency can be improved by having access to AI (Artificial Intelligence) based contours. MATERIALS AND METHODS: The 63 patients in the DAHANCA 35 pilot trial had a CT from the local DAHANCA centre and one from the proton centre. A nationally validated convolutional neural network, based on nnU-Net, was used to contour OARs on both scans for each patient. Using deformable image registration, local AI and oncologist contours were transferred to the proton centre scans for comparison. Consistency was calculated with the Dice Similarity Coefficient (DSC) and Mean Surface Distance (MSD), comparing contours from AI to AI and oncologist to oncologist, respectively. Two NTCP models were applied to calculate NTCP for xerostomia and dysphagia. RESULTS: The AI contours showed significantly better consistency than the contours by oncologists. The median and interquartile range of DSC was 0.85 [0.78 - 0.90] and 0.68 [0.51 - 0.80] for AI and oncologist contours, respectively. The median and interquartile range of MSD was 0.9 mm [0.7 - 1.1] mm and 1.9 mm [1.5 - 2.6] mm for AI and oncologist contours, respectively. There was no significant difference in ΔNTCP. CONCLUSIONS: The study showed that OAR contours made by the AI algorithm were more consistent than those made by oncologists. No significant impact on the ΔNTCP calculations could be discerned.

Technical note: Temporal and thermal stability of optical response for silicone-based 3D radiochromic dosimeters.

BACKGROUND: Radiochromic silicone-based dosimeters are flexible 3D dosimeters, which at appropriate concentration of leucomalachite green (LMG) and curing agent are dose-rate independent for clinical photon beams. However, their dose response is based on chemical processes that can be influenced by temporal and thermal conditions, impacting measurement stability. PURPOSE: The aim of this study was to investigate the temporal stability of the dose response of radiochromic dosimeters for different curing times and post-irradiation storage temperatures. METHODS: Six cylindrical dosimeters (5 cm diameter, 5 cm length) were produced in a single batch and separated into two groups that were irradiated 72 and 118 h after production. The same photon plan, consisting of two 10 × 1.6 cm2 opposing fields, was delivered to all dosimeters. After irradiation, the dosimeters were separated into three groups, stored at 5°C, 15°C, and 20°C, and read out for five consecutive days. RESULTS: Storage temperature influenced the measurement stability, and changes in the optical response with time differed between irradiated and non-irradiated parts of the dosimeters. The relative change between signal and background was greater than 10% for all measurements performed 24 h or more after irradiation, except for dosimeters stored at 5°C, which changed by 2%-5% after 24 h. The dosimeter temporal stability was not influenced by curing time. CONCLUSIONS: For room temperature storage (15°C and 20°C), readout should take place as soon as possible after irradiation since the background color increased rapidly for both curing times (72 and 118 h), whereas the dosimeters are stored at 5°C, readout can be performed up to 24 h after.

Empirical quenching correction in radiochromic silicone-based three-dimensional dosimetry of spot-scanning proton therapy.

BACKGROUND AND PURPOSE: Three-dimensional dosimetry of proton therapy (PT) with chemical dosimeters is challenged by signal quenching, which is a lower dose-response in regions with high ionization density due to high linear-energy-transfer (LET) and dose-rate. This study aimed to assess the viability of an empirical correction model for 3D radiochromic silicone-based dosimeters irradiated with spot-scanning PT, by parametrizing its LET and dose-rate dependency. MATERIALS AND METHODS: Ten cylindrical radiochromic dosimeters (Ø50 and Ø75 mm) were produced in-house, and irradiated with different spot-scanning proton beam configurations and machine-set dose rates ranging from 56 to 145 Gy/min. Beams with incident energies of 75, 95 and 120 MeV, a spread-out Bragg peak and a plan optimized to an irregular target volume were included. Five of the dosimeters, irradiated with 120 MeV beams, were used to estimate the quenching correction factors. Monte Carlo simulations were used to obtain dose and dose-averaged-LET (LETd) maps. Additionally, a local dose-rate map was estimated, using the simulated dose maps and the machine-set dose-rate information retrieved from the irradiation log-files. Finally, the correction factor was estimated as a function of LETd and local dose-rate and tested on the different fields. RESULTS: Gamma-pass-rates of the corrected measurements were >94% using a 3%-3 mm gamma analysis and >88% using 2%-2 mm, with a dose deviation of <5.6 ± 1.8%. Larger dosimeters showed a 20% systematic increase in dose-response, but the same quenching in signal when compared to the smaller dosimeters. CONCLUSION: The quenching correction model was valid for different dosimeter sizes to obtain relative dosimetric maps of complex dose distributions in PT.

Technical Note: Improving proton stopping power ratio determination for a deformable silicone-based 3D dosimeter using dual energy CT.

PURPOSE: The aim of this study was to investigate whether the stopping power ratio (SPR) of a deformable, silicone-based 3D dosimeter could be determined more accurately using dual energy (DE) CT compared to using conventional methods based on single energy (SE) CT. The use of SECT combined with the stoichiometric calibration method was therefore compared to DECT-based determination. METHODS: The SPR of the dosimeter was estimated based on its Hounsfield units (HUs) in both a SECT image and a DECT image set. The stoichiometric calibration method was used for converting the HU in the SECT image to a SPR value for the dosimeter while two published SPR calibration methods for dual energy were applied on the DECT images. Finally, the SPR of the dosimeter was measured in a 60 MeV proton by quantifying the range difference with and without the dosimeter in the beam path. RESULTS: The SPR determined from SECT and the stoichiometric method was 1.10, compared to 1.01 with both DECT calibration methods. The measured SPR for the dosimeter material was 0.97. CONCLUSIONS: The SPR of the dosimeter was overestimated by 13% using the stoichiometric method and by 3% when using DECT. If the stoichiometric method should be applied for the dosimeter, the HU of the dosimeter must be manually changed in the treatment planning system in order to give a correct SPR estimate. Using a wrong SPR value will cause differences between the calculated and the delivered treatment plans.