Research

Research in the Yang Advanced Adaptive Radiation Therapy (A²RT) Laboratory includes the following:

Deformable registration is the enabling technique for adaptive radiotherapy. Our lab focuses on the algorithm development to address special difficult cases in deformable registration to improve registration accuracy, including non-correspondence, large deformation, lung sliding cases, etc. We also develop respiratory motion models using deformable registration to investigate motion control strategy, treatment margin recipe and dosimetric effect. In addition, we designed an in-house phantom for end-to-end evaluation of registration accuracy. Currently, we are focusing on artificial intelligence (AI) based algorithm development for CT/MR deformable registration and quality assurance of deformable registration for online adaptive radiotherapy.

Modern radiation therapy techniques require clinical specialists accurately define the targets and the concerned organs-at-risks for radiation treatment planning. Traditional manual contouring is labor-intensive and subject to intra- and inter-observer variations that are caused by different clinical experience, training of the specialists and quality of available medical images. Over the past several years, our lab have been dedicated to the development of multi-atlas segmentation to facilitate normal tissue autocontouring during treatment planning. We have created a software tool, Multi-Atlas Contouring Service (MACS), which interfaces with the Pinnacle and RayStation treatment planning system, for clinical use of auto-contouring. Currently, we are focusing on developing AI-based algorithms for autocontouring gross tumor volumes from multi-modality images (PET/CT/MR) and MR-based autocontouring for online adaptive planning.

Recent advances in the integration of MRI with a linear accelerator (MR-Linac) have enabled MR-guided radiotherapy (MRgRT). Our group has been focusing on the translational clinical research of treating patients with Elekta Unity MR-Linac, including clinical feasibility, dose accumulation and online planning strategy, etc. Currently we are working toward automating the online adaptive planning, including auto-selection of adaptation approaches, auto-contouring, auto-verification of contouring accuracy, quality assurance of online adaptive plan and synthetic CT for MR-based treatment planning.

Quantitative imaging biomarkers or radiomics features are quantitative descriptors that reflect textural variations in image intensity, shape, size or volume to offer information on tumor phenotype. Our group investigated the uncertainties of the imaging biomarkers due to imaging scanners, imaging protocols, reconstruction algorithms and tumor segmentations. We have explored the possibility of using MRI delta radiomics together with machine learning approaches for predictive modeling to distinguish radiation necrosis from tumor progression after Gamma Knife radiosurgery for brain metastases. Currently we are exploring the use of DWI to predict MR-Linac treatment outcome for prostate SBRT.