Research

Project 1: Imaging-based biomarkers

We have been developing image-based biomarkers to improve prognostication and prediction of therapy response in cancer. Our approach incorporates quantitative radiomic profiling of the tumor and intratumoral subregions, as well as surrounding parenchyma. We are also exploring the biological basis of clinically relevant imaging phenotypes by correlating with molecular data. Importantly, we have shown that the newly identified imaging markers complement established biomarkers and their integration further improves prediction accuracy. We have been conducting imaging biomarker studies on breast, lung, brain, head and neck cancers, using a variety of modalities including PET, CT, and MRI (Cui et al. Radiology, 2016, Wu et al. Radiology, 2016, Wu et al. Radiology, 2017, Wu et al. Clinical Cancer Research, 2017, Wu et al. Radiology, 2018).

Project 2: Molecularly-based biomarkers    

In addition to our work on imaging, we are also developing molecular biomarkers to predict therapy response and prognosis in several cancer types. Recently we have developed an individualized immune-based gene expression signature that predicts survival of patients with non-small cell lung cancer and validated it in large cohorts of over 2,000 patients (Li et al. JAMA Oncology, 2017). In another work, we developed an integrated radiosensitivity and immune gene signature that predicts which patients are most likely to benefit from radiotherapy in breast cancer (Cui et al. Clinical Cancer Research, 2018). Additionally, we are interested in early detection of aggressive cancers through analyses of the genetic profiles.

Project 3: MRI-based radiation treatment planning

MRI has important advantages over the current gold standard – CT for radiation treatment planning, including improved target delineation. However, the lack of electron density information in MRI has been a major technical hurdle for its clinical adoption. We developed a unifying Bayesian approach that combines both geometry and intensity information for more reliable electron density mapping using MRI (Gudur et al. PMB, 2014). In ongoing work we are further improving the accuracy and robustness of this new technique.

Acknowledgment

Our research has been supported by multiple grants from the National Cancer Institute (K99/R00 CA166186, R01 CA193730, R01 CA222512) at the NIH.