Ionizing radiation acoustic and ultrasound dual-modality imaging for visualization of dose on anatomical structures during radiotherapy

Abstract

The aim of this study is to visualize the radiation dose on anatomical structures during radiation therapy (RT) by mapping radiation dose deposition and tracking anatomical structures simultaneously. A dual-modality volumetric imaging system, which combines ionizing radiation acoustic imaging (iRAI) and ultrasound (US) imaging, was developed to provide dose deposition and anatomical information in real-time during RT. The performance of the proposed system was first evaluated via experiments on tissue-mimicking phantoms driven by a custom motion stage. By using US imaging to correct the position of anatomical structures, the dose mapping accuracy of the system increased by up to 0.51 in structural similarity index measure (SSIM) and 74.60 % in Gamma passing rate (GPR) compared to standalone iRAI. A subsequent study on a rabbit model in vivo further confirmed the capability of the system in mapping of the radiation dose deposition in the target tissue as well as its change caused by the motion mainly due to the animal breath. These findings demonstrate that this first-of-its-kind dual-modality volumetric imaging system can provide volumetric dose-on-anatomy information during RT. After further validation in clinic, this technique holds potential for enhancing RT outcomes by ensuring accurate alignment between the planned radiation beams, the target, and surrounding organs at risk.

Keywords: Dose mapping on anatomical structures; Ionizing radiation acoustic imaging (iRAI); Radiation therapy precision; Ultrasound (US) imaging.

Fig. 1

Schematic and operating principle of the iRAI and US dual-modality imaging system. a, Schematic of the iRAI and US dual-modality imaging system for in vivo dose mapping on anatomical structure. b, Specifications of the iRAI 2D MAT and the US 2D MAT. c, Timing sequence of the dual-modality imaging system. d, The measured spatial resolution of the iRAI system as a function of the depth from the probe surface. Elevational resolution is the same as the lateral. e, The measured spatial resolutions of the US imaging system as a function of the depth from the probe surface. Elevational resolution is the same as the lateral.

Fig. 2

Study on a respiratory motion-mimicking phantom. a, Experimental setup for the study on the tissue-mimicking phantom. b, Detailed illustration of the phantom study, including lateral and axial motions of the targeted block of lard. An x-y-z coordinate system is marked. c, Schematic illustration of the 2D MAT for US imaging of motion, the 2D MAT for iRAI of dose deposition, and the phantom embedded with a lard block. The schematic shows the misalignment between the temporally deposited radiation dose and the lard block, indicating the spatial shift. The radiation beam is directed perpendicularly to the plane of the page. d, x-z cross-sectional view of a representative iRAI image of a 2 cm × 2 cm square radiation beam. e, Original signal extracted from the white dashed line in d, along with the corresponding post-enveloped signal. f, Post-processed iRAI image of the square beam.

Fig. 3

3D US imaging of the phantom with a lard block at various frame rates. a-c, Illustration of dynamic focusing for the US 2D matrix array transducer, with the number of focus positions set to 16, 9, and 4, respectively. d-f, x-z cross-sectional US images of the lard block at frame rates of 2.5, 4, and 10 Hz, corresponding to 16, 9, and 4 focusing positions.

Fig. 4

Temporal iRAI dose on phantom structure images across the isocenter. a, US, iRAI temporal dose, and fusion image series shown as x-z cross-sectional views across the isocenter acquired from lateral motion experiment. b, US, iRAI temporal dose, and fusion image series shown as x-z cross-sectional views across the isocenter acquired from axial motion experiment. Both the image sets in a and b were captured during a period of 2 s, with US and iRAI images acquired every 0.25 s. All images are displayed with intensities normalized to the range of 0–1, with US images shown in yellow pseudo color and iRAI images in red pseudo color.

Fig. 5

iRAI dose mapping using US motion correction. a, x-z cross-sectional iRAI images of temporal dose accumulation across the isocenter relative to the lard block over a 2-second delivery period, with frames shown every 0.25 s. Pixel values were normalized from 0–1. The green dashed boxes indicate the lard block position. b, Examples of X-ray dose accumulation recorded on radiochromic films, alongside film dose maps and corresponding x-z cross-sectional view of iRAI dose maps across the isocenter from the phantom experiment with motion along the lateral direction. Film dose maps were extracted from a 2 cm × 6 cm rectangular area. The four groups of figures represent different motion modes with speed of 2 cm/s and moving ranges of 1, 2, 3, and 4 cm, respectively. Pixel intensities were normalized in both film and iRAI maps. c, Examples of X-ray dose accumulation recorded on radiochromic films, alongside film dose maps and corresponding z-x cross-sectional view of iRAI dose maps across the isocenter from the phantom experiment with motion along the axial direction. The four groups of figures represent different motion modes with speed of 1 cm/s and moving ranges of 0.5, 1, 1.5, and 2 cm, respectively. d, Schematic of the dose map with a dashed white line across the center, indicating the location where the dose profile is extracted. e and f, Intensity profiles derived from the film and iRAI dose maps shown in b and c, corresponding to the line marked in d, over lengths of 8 cm and 6 cm, respectively. Dashed lines represent film dose profiles, while solid lines represent iRAI dose profiles.

Fig. 6

Volumetric multi-modality imaging of dose-on-anatomy in a rabbit model in vivo. a, Experimental setup for the in vivo rabbit study in the radiotherapy room. b and c, Sagittal and coronal views of the rabbit treatment plan visualized by CT, respectively. d, Time-dependent multi-modality image series, with the first row displaying sagittal cross-sectional views and the second row displaying coronal cross-sectional views. iRAI dose deposition is shown in red pseudo-color, US image is shown in yellow pseudo-color, and the liver boundary, extracted from CT scans, is outlined with a light blue line. The color bars on the right indicate normalized relative amplitude or dose, respectively. e, 3D multi-modality fusion images combining CT, US, and iRAI at the 1st, 10th, and 19th seconds of radiation dose delivery, respectively. Tissue motion, captured by US imaging, and dose variation, represented by iRAI, were visualized relative to the initial anatomical structures from CT scans.