Research
Dr. Caroline Chung leads an imaging computational laboratory within the department of Radiation Oncology at MD Anderson Cancer Center. The Chung Lab's major research focus is to develop quantitative imaging pipelines and predictive tools to be used in:
- Tumor response assessment,
- Treatment-related toxicity, and
- Personalization of radiotherapy and multimodal treatment.
In addition, the lab is working on the standardization of collection and nomenclature of images to facilitate meaningful measurement and interpretation of imaging biomarkers across departments and institutions to support efforts aligned with the Institute for Data Science in Oncology (IDSO).
Quantitative imaging research is a key component to enabling and guiding personalized oncological patient care. The Chung Lab has an additional role in supporting the Tumor Measurement Initiative (TMI), which aims to build an institutional platform to support standardized, automated, quantitative imaging-based tumor measurement across each patient’s journey to advance multidisciplinary, data-driven, high-precision cancer treatment.
Retrospective Determination of Prognostic Factors in BRAF-V600E Melanoma Brain Metastases
BRAF-mutant melanoma brain metastases (MBM) have poor prognosis with overall survival for these patients averaging 4 months from MBM diagnosis. Despite this poor prognosis, the disease is historically understudied and there is little evidence arguing for a standardized treatment plan across the field. This study aims to retrospectively identify clinical and treatment-related variables that may predict differences in overall survival for patients with BRAF-mutant MBM that could help inform a data-driven standard of care.
Probabilistic Tumor Segmentation for assigning a probability level to a tumor and its immediate surrounding area
The goal of this research is to apply deep learning methods to generate a probability map that will distinguish tumors and the affected tissue around them from the rest of the brain. The ability to identify not only the main tumor body but also the surrounding tissue that has been infiltrated by malignant cells, along with tumor tracking over time is a major problem in maximizing the efficacy of cancer treatment. In addition, the use of deep learning and other machine learning techniques to segment brain tumors and speed up diagnosis has been hampered by the complex nature and shape of these tumors.
Owing to the highly varied nature of these tumors, a one-size-fits-all approach does not work and a more flexible method is required. We propose a method where an individual probability map will be produced for each size and tumor type, then at the final stage of calculation these individual maps will be combined into one. These probability maps will also allow for tumor tracking and size change analysis over time.
Post-op treatment planning study
A common course of treatment for patients with GBM is surgery for tumor resection followed by radiation and chemotherapy. In some cases, usually due to the cost, time, or lack of resources (MRIs), a patient will receive only a CT at the time of simulation for radiation therapy planning. Then, the post-op (or otherwise most recent) MRI will be registered to the planning CT for target and OAR delineation. The adequacy of this approach relies on the assumption that the tissue has not changed substantially between surgery and RT-Sim. This project will assess the adequacy of this assumption by retrospectively planning RT on post-op images of patients in the adaptive GBM trial.
NASA Cognitive Protocol
NASA has identified several gaps in the Human Research Roadmap concerning the functional consequences of unavoidable exposure of the brain to ionizing radiation during spaceflights. In this prospective study of patients receiving radiotherapy to tumors in the head and neck/skull base region, biofluid and imaging measures are being collected serially to identify predictive biomarkers of neurocognitive toxicity from brain irradiation. Brain connectomics is a tool being used to predict neurocognitive decline. Structural connectivity and functional connectivity are respectively derived from diffusion tensor imaging (DTI) that maps white matter tracts and resting-state functional MRI (rs-fMRI) that identifies temporally correlated grey matter activity. This study aims to determine the associations between radiation-induced neurocognitive impairments and dose to network hubs through quantitative connectivity changes.
Areas of Research
Data Science
Imaging Biomarkers
Machine Learning
Neural Oncology
Personalized Radiotherapy
Quantitative Imaging
Radiation Toxicity
Stereotactic Radiosurgery
Clinical Trials
For the GBM Adaptive trial, patients received multimodality MR imaging periodically throughout their course of radiation therapy. This study will perform DIR on the follow-up images to track the brain tissue as it shifts from planning to the end of therapy and determine the accumulated dose and look for a correlation with local recurrence and necrosis of healthy tissue.
Personalization of Glioma Treatment via Imaging-Informed Mechanistic Models of Tumor Progression
The goal of this project is to develop an experimental-computational framework to predict and spatially map biologically relevant, treatment-resistant regions of tumor to enable personalized timely adaptive RT treatment in patients receiving RT for high grade glioma. This is done by employing multi-parametric MRI as well as weekly non-enhanced MR images to calibrate a predictive biologically based mathematical model of tumor growth and response to treatment. The calibrated model parameters are then used to forecast individual response at future imaging visits and help create patient-specific adaptive RT changes.
Mathematical modeling holds promise to better understand tumor progression and resistance for patients with high grade gliomas receiving radiation treatment. In this project, we leverage the power of mathematical modeling and multi-parametric MRI to make biologically based predictions of tumor growth and response to treatment. We aim to calibrate a family of mathematical models to forecast individual response before the end of radiation in order to recommend patient-specific adaptations to the planned radiation treatment.
Quantitative Imaging Biomarkers of Immune Response in Melanoma Brain Metastases
Immunotherapy and targeted therapy for metastatic melanoma have substantially changed the prognosis for patients with brain metastases due to both promising intracranial responses and improved extracranial disease control. Particularly in the brain, where tissue is often difficult to acquire, non-invasive approaches of characterizing tumor immunophenotype and assessing treatment response are greatly needed in the age of personalized medicine. In this project we will develop and validate quantitative imaging biomarkers of immunophenotype and immune response in patients with melanoma brain metastases treated with immunotherapy +/- radiotherapy using conventional and advanced magnetic resonance imaging (MRI).
While immunotherapy and targeted therapy have significantly improved outcomes for patients with metastatic melanoma to the brain, providing truly personalized medicine to these patients remains challenging. Acquiring tissue from the brain is invasive, but currently the standard method to characterize immunophenotype and treatment response. Developing non-invasive methods to understand these variables is a big step toward improving outcomes for these patients. This project aims to develop and validate quantitative imaging biomarkers of immunophenotype and response in patients with melanoma brain metastases who are treated with immunotherapy with or without radiation, using conventional and advanced MR imaging.
This trial studies whether a customized video intervention can help to reduce anxiety in brain cancer patients undergoing radiation treatment and their caregivers. A customized neuro-imaging referenced symptom video that describes symptoms and side effects specific to the patients' tumor may result in an early and sustained reduction in anxiety and distress during and after radiation treatment, thereby improving quality of life.
This trial studies how well nTMS works in planning for stereotactic radiosurgery in patients with brain metastases in the motor cortex. Stereotactic radiosurgery is a type of radiation therapy that delivers high doses of radiation, which can sometimes lead to damage occurring to the brain and surrounding areas. The motor cortex (the part of the nervous system that controls muscle movement), however, currently has no radiation dose limit. nTMS is a non-invasive tool that uses sensors on a patient's muscle to trace the location in their brain that controls that muscle and is currently used by doctors to decide where to operate so as not to damage the motor nerves. nTMS may effectively help plan radiation treatment using SRS and help doctors decide on how much radiation can be used on motor nerves.