Research Overview
My research focuses on the development and application of innovative imaging and medical image analysis methods to improve the diagnosis, treatment and understanding of cancer.
I am interested in three broad research areas. The first area is the development of robust and accurate image registration algorithms for image guidance in radiation therapy, where accurate and reproducible patient setup during irradiation is essential in order to ensure that the treatment is delivered as planned. An example of this research is the development of registration algorithms to estimate patient setup errors, and hence reduce targeting uncertainties, during external beam radiation therapy for prostate cancer through 2D-3D registration of 2D MV low-contrast portal images or 2D kV radiographs to 3D kV planning CT or 3D cone-beam CT images. Inherent to the methods I developed is an understanding of the physics of CT and x-ray images, particularly the spectral and statistical characteristics of signal and noise in these images.
A possible consequence of radiation therapy is the development of complications due to exposure of healthy tissues to radiation. My second area of research involves the identification of factors that contribute to late rectal toxicity in the treatment of localized prostate cancer with IMRT using mathematical and quantitative image analysis techniques. Identification of such factors could allow us to predict the risk of a patient developing toxicity following treatment and also allow the formulation of clinical guidelines to reduce the risk of complications arising. In this work, I have extended upon conventional methods for dosimetric analysis to account for the spatial and shape characteristics of the dose distributions in addition to the amount of tissue irradiated to a given dose, and their influence on the incidence of toxicity. Such methods may provide insight, for example, into the spatial sensitivity of the organ irradiated to radiation dose or whether the presence of lower dose regions between regions of high dose may help reduce radiation damage.
My third area of interest is the precise localization, staging and grading of tumors, and the modeling of tumor progression using advanced segmentation methods and imaging techniques for imaging at the cellular and molecular level. While image-guidance in the diagnosis and treatment of cancer is useful, accurate localization and delineation of cancerous cells is also essential to achieving better local tumor control. Accurate characterization of the tumorous volume, however, remains rather challenging. Progress in these areas would not only help provide better diagnosis and treatment but also have a more long-term contribution to our understanding of the biology of cancer. Potential projects include the delineation and grading of tumors with poorly defined boundaries such as brain gliomas from MRI images or MRS images and the use of quantitative imaging biomarkers to monitor glioma status.
I have also initiated a monthly research seminar series in biomedical imaging at Weill Cornell Medical College. The series is designed to highlight the work of scientists developing and applying new techniques and approaches in medical imaging and analysis, and is documented at the following website:
http://www.weill.cornell.edu/research/cbic/research/seminars.html
