Breast SBRT

Tsuicheng Chiu from UT Southwestern Medical Center described a prototype in-room 3D volumetric ultrasound guidance system for stereotactic body radiotherapy (SBRT) of the prone breast. Chiu began by detailing the challenges in delivering prone breast SBRT, including: the need to immobilize the breast, which usually involves compression and tissue deformation; and accurate patient positioning, where markers are often used, despite exhibiting some mobility in soft tissue. "We propose the use of ultrasound as an image-guidance technique to overcome those challenges," he said.

Chiu and colleagues have designed a non-touch scanning system that is placed beneath a prone breast board. The breast is immobilized in a patient-specific holder (based on the patient's CT scan and made from liquid rubber) that does not deform the tissue. The holder fits into a portable ultrasound scanner within a water bath. After the scan, the ultrasound system is removed, but the holder remains to keep the breast in the same position during treatment.

The ultrasound system performs a rotational scan (rotating the water tank and transducer together), acquiring images at 5° steps. After each planar scan (72 raw images), the transducer is elevated to the next plane. A 3D ultrasound image is then constructed using a supercompounding-technique-based volumetric reconstruction algorithm.

The researchers evaluated their prototype system using a breast phantom with 10 embedded markers. A single raw image exhibited artefacts and shadows. "However, the final reconstructed images showed higher contrast and smoothed speckle," noted Chiu. They also compared the compound ultrasound images to CT (which could not visualize the markers) and MR images.

Results showed good geometric agreement between MR and ultrasound images, indicating that the proposed guidance system can provide precise and accurate patient set-up. MR and ultrasound images of the various embedded markers differed by 1.0–2.8 mm in diameter; measured distances between adjacent markers differed by 0.5–0.9 mm. "The ultrasound images are slightly larger because ultrasound has a different response at boundaries between different materials," Chiu explained. "But if we know about this, then we can compensate for this disadvantage."

Chiu concluded that the ultrasound guidance system provides high-quality volumetric images with high soft-tissue contrast. He noted that the ability to easily image patients before every fraction enables treatment response monitoring by visualizing changes in the target.

Responding to an audience question, Chiu explained that the prototype system is semi-automatic and takes about one minute to scan each plane. "Currently we only have a 2D array," he said. "But once we use a 3D array, the system will scan the entire breast in about one minute, because it won't need to move the transducer up and down. This will save a lot of time."

Soft-tissue imaging

Speaking in the same conference session, Justin Sick from Purdue University presented a 3D ultrasound platform for image-guided radiation treatment of transitional cell carcinoma, a cancer that typically occurs in the urinary system. The goal, he explained, is to develop the tools necessary for clinics to implement intra-fraction motion management into the radiotherapy suite.

The proposed system comprises an ultrasound transducer that slots onto rails that are mountable onto a radiotherapy treatment couch. The angle of the transducer can be adjusted in 15° increments and its position can be fine-tuned in 10 µm increments – providing a wide range of imaging positions. The Purdue University team also developed methods to relate the CT, 3D-ultrasound and linac coordinate systems.

Sick and colleagues tested the accuracy of the system by setting the transducer at the isocentre, moving it slightly, determining the shift and moving the couch accordingly. Average errors between predicted and actual couch positions were 0.18±0.85, –0.001±0.99 and 0.1±0.72 mm, in the lateral, vertical and longitudinal directions, respectively. Sick noted that these values meet the ±1 mm accuracy requirements of TG-179 for CT-based guidance technologies.

To provide a means for quality assurance of the ultrasound platform, the researchers created an agar/glycerine-based phantom with density and speed-of-sound values mimicking published values for abdominal soft tissues (water, fat, skin, muscle, urine). They imaged the phantom under various rotations and translations relative to the CT and ultrasound coordinate systems.

The volumetric ultrasound images were imported (as a 3D matrix) into the Eclipse treatment planning system, where they could be co-registered with the simulation CT. The co-registered ultrasound-to-ultrasound and ultrasound-to-CT images were then used to re-align the phantom relative to the linac.

The transformation matrix components displayed an average positioning error of 1.99±1.42 mm. "The rotational discrepancies between ultrasound and CT, and ultrasound and ultrasound were all very small," said Sick, noting that translational errors were larger, but acceptable if the rotation remained below 5°. He concluded that ultrasound guidance can provide equivalent performance to kV and MV techniques currently used for positioning prostate patients.

Related stories

• Ultrasound can guide MLC tracking
• Ultrasound eases prostate motion tracking
• Ultrasound tackles radiotherapy guidance
• Ultrasound guidance for prostate IGRT