Collaborative Projects

Multimodal Prostate Cancer Detection
In collaboration with Dr. Abigail Haka, Cornell Medical College, we are developing a more accurate and less invasive procedure for the diagnosis of prostate cancer by exploring the ability of Raman spectroscopy to differentiate benign and malignant lesions and to provide information concerning prognosis and lesion grade on ex vivo tissue samples.

Prostate cancer is the second most common cancer in American men. In the United States, an estimated 192,000 cases were diagnosed in 2009 and 27,000 deaths occurred. Patients typically present with elevated serum prostate specific antigen (PSA), genitourinary symptoms, or an abnormal digital rectal exam. However, these symptoms are not uniquely correlated with the probability of prostate cancer. As such, a histologic examination of the tissue is required to establish the diagnosis of prostate cancer. This is typically accomplished via transurethral biopsy of the prostate (TURP) in which 6-12 cores of tissue are removed (more for patients with prostatic hypertrophy). Sites to be biopsied are often determined by findings on digital rectal examination or ultrasonography. However, in many men these tests are normal and blind biopsies are carried out, based on a generalized map of the prostate. A substantial number of men with negative initial biopsy but persistently high serum PSA will have cancer diagnosed on subsequent biopsies. As a consequence of the limitations of current techniques, each year a large number of prostate biopsies are performed on lesions ultimately diagnosed as benign. The complete diagnostic process, from start to finish, may take months and include multiple biopsies. The desirability of reducing the number of benign biopsies performed, patient trauma, time delay and the high medical costs associated with biopsy has motivated researchers to explore minimally invasive optical methods for diagnosing prostate lesions.

Our studies consist of data acquisition from a large number of patients and pathologies using samples from intra-surgical biopsies and freshly excised prostate. Additionally, specimens from the Cornell tissue bank are utilized. Unlike many forms of optical spectroscopy, the Raman signal is largely unaffected by tissue freeze thaw. The tissue bank provides the ideal opportunity to develop not only a diagnostic, but also a prognostic (the likelihood of the cancer to be indolent vs. aggressive) algorithm. In this context, the banked specimens are advantageous in that prognostic data on these patients are available several years after the specimens were obtained. We use the data to test the ability of Raman spectroscopy to differentiate benign and malignant lesions and to provide information concerning prognosis and lesion grade. Data is analyzed with principal component analysis and logistic regression for initial algorithm development. However, once the clinical Raman system is modified to allow confocal data acquisition, a morphological model, similar to those previously utilized in breast and artery, will be developed for prostate tissue. The model will then be used to provide information about tissue composition as well as diagnoses.