Towards quantitative ionizing radiation acoustic imaging (iRAI) for radiation dose measurement: Validation from simulations to experiments

Background

In clinical radiation therapy (RT), accurately quantifying the delivered radiation dose to the targeted tumors and surrounding tissues is essential for evaluating treatment outcomes. Ionizing radiation acoustic imaging (iRAI), a novel passive and non-invasive imaging technique, has the potential to provide real-time in vivo radiation dose mapping during RT. However, current iRAI technology does not account for spatial variations in the detection sensitivity of the ultrasound transducer used to capture the iRAI signals, leading to significant errors in dose mapping.

Purpose

This paper presents the first detection sensitivity-compensated quantitative iRAI approach for measuring deposited radiation dose, aiming at improving dose mapping accuracy.

Methods

Detection sensitivity maps for the 2D matrix array transducer (MAT) were generated through both computational studies and experimental measurements. First, the Field II MATLAB toolbox was used to simulate the acoustic fields generated by the 2D MAT at various focal angles in the region of interest. Second, the prototype 2D MAT was applied to experimentally measure the acoustic signals generated by pulsed laser point sources distributed throughout the same volume as in the simulation. Then, in vitro experiments were conducted using homogeneous soft-tissue phantoms, where x-ray beams with square fields and a C-shaped treatment plan were separately delivered via a clinical linear accelerator (LINAC). Additionally, the propagation of acoustic waves induced by the x-ray beams with square fields was simulated using the K-Wave MATLAB toolbox. Correction factors derived from both the simulated and experimental sensitivity maps were applied to compensate for sensitivity-induced discrepancies in the iRAI reconstruction results. Dose distributions in uncompensated and sensitivity-compensated iRAI volumetric images were compared across various beam positions and field sizes. The agreement between the iRAI images and the treatment plan was quantitatively evaluated using structural similarity index measure (SSIM) and gamma index analysis.

Results

The experimental results, including the detection sensitivity map and iRAI measurements of x-ray beams with square fields, showed strong agreement with the corresponding simulated outcomes. Following compensation, the relative amplitudes of all iRAI images for beams targeting different positions converged toward 1. The compensated iRAI images revealed greater agreement with the treatment plan in dose distribution, compared to the pre-compensation images. This improvement was further supported by global gamma index analysis, which showed an increase in the 5%/5 mm dose difference (DD) /distance-to-agreement (DTA) passing rate from 56.86% to 78.24% after compensation, indicating improved accuracy in reconstructing the dose distribution.

Conclusions

This study demonstrated that addressing inhomogeneities in transducer detection sensitivity significantly enhances the accuracy of radiation dose mapping by iRAI.