Online image data allows for adaptive therapy response
Metastatic disease changes size and shape when undergoing radiation treatment. In addition, soft tissue and organs shift position during therapy. For radiation oncologists, adapting a treatment plan for these deformations is a challenge.

Solutions to this problem range from shifting the treatment table during therapy to the time-consuming task of adapting the treatment plan at each fraction. A group of researchers from the department of radiation oncology at Emory University School of Medicine in Atlanta has developed a software tool that allows physicians to adapt their treatment plan on a daily basis to account for interfraction organ motion.

“Two major obstacles of adaptive radiation therapy (ART) are manual anatomical segmentation and dose optimization/calculation,” the scientists wrote in one of two scientific poster presentations about the technology at the recent American Society for Therapeutic Radiology and Oncology (ASTRO) conference in Los Angeles. “Currently, these tasks cannot be completed in a short time period while the patient lies on the treatment table.”

The team has developed an automated online adaption method for segmenting prostate anatomy and planning dynamic arc-based therapy delivery for prostate cancer patients.

Their system replaces manual segmentation with automated atlas-based segmentation based on a level-set approach. They utilize an algorithm that permits transfer of segmented contours between the planning CT and cone beam CT (CBCT) by deformable image registration.

“Our method involves observing soft-tissue deformation in pre-treatment CBCT acquired daily via a LINAC-mounted kilovoltage imaging system in the treatment delivery room,” they wrote. “First, a rigid image registration based on the prostate only gives the best match achievable using couch translations alone. Then, the residual anatomical deformations are obtained using a level-set segmentation method that deforms the structures delineated in the planning CT to the anatomy observed in the daily CBCT images.”

The group tested the software on 10 patients undergoing CBCT for prostate therapy at their institution.

“A rigid registration of soft-tissue takes only 20 seconds, with an additional 90 seconds needed for the deformable segmentation and 20 seconds needed for the dose-volume histogram (DVH) recalculation,” they wrote. “The software provides recommendations on the utility of re-optimization based on the observed range of deformations for the prostate and whether the re-computed DVH meets the original coverage criteria. However, at the same time, it interactively displays segmentation results overlaid on the CBCT and re-computed DVHs.”

The scientists put the tool to use in the delivery of dynamic conformal arc-based therapy instead of inverse planning. According to the researchers, arc-based therapy is an alternative to intensity modulated radiation therapy (IMRT) that uses multiple geometrically-optimized beams instead of a few intensity-modulated beams. This makes the dose-optimization process much faster as it involves only automated geometrical shaping of the multi-leaf collimator to the planning target volume (PTV) margins.

“We hypothesize that even if the dose distribution produced by the dynamic arc therapy may be qualitatively worse than the one obtained by inverse planning, the gain obtained from being able to reduce the target margin by adapting the treatment plan to daily changes in anatomy overcomes the reduction in dose conformality associated with using arc-based therapy,” they wrote.

With their advanced visualization and planning application, the team was able to obtain a dynamic conformal arc-based therapy plan in 80 seconds.

“In individual fractions, the dose irradiating 60 percent of the volume of the bladder and rectum was reduced from 46.1 percent and 31.4 percent of the prescription dose to 12.7 percent and 8.4 percent while keeping the same target coverage,” they noted.

The online adaptive software for segmenting prostate anatomy is able to provide clinical guidance on whether arc optimization is need for each treatment fraction on a daily basis while the patient is on the treatment table. In addition, the tool allows for the delivery of what may be a more effective treatment method.

“In summary, significant improvement in the dose to the critical structures was observed in the online adaptive method as compared to IMRT without interfraction motion management,” the researchers wrote. “This was attributed to reducing the PTV margin of 10mm, which accounts for organ motion and patient setup uncertainties, to 5mm to account for automated segmentation inaccuracies.”