Combining visual and infrared information for diagnosis and prognosis in prostate cancer
Prostate cancer is the most common cancer in men, and strongly correlated with age. Given the rapidly aging population, the number of people diagnosed with cancer is going to rise. A concomitant problem is the number of biopsies that will have to be taken and analyzed, particularly given recent calls for the introduction of increased routine screening. This is going to put a severe strain of all parts of the health service but particularly NHS pathology departments.
At present biopsy tissue samples are analyzed under a microscope by a trained pathologist. A diagnosis is made after examining multiples tissue sections looking for specific changes in tissue architecture and is aided by the use of histochemical staining This is a slow time consuming process and inevitably subject to a degree of human variability. There is a strong driving force therefore to develop new methods of analysis that can be automated and are objective.
Recently, infrared hyperspectral imaging coupled with data analytic tools has been shown to reveal the full chemical composition, pixel by pixel, of the same sample used by the pathologist [1, 2]. Thus the distribution of constituents such as proteins, carbohydrates, DNA, etc. - the chemistry of the cancer - can be found.
The goal of this project is discover a way to fuse the shape information used by the pathologist, with the spectral information provided by the spectroscopist, to develop a new, objective method of cancer tissue classification that will be able to automate diagnosis and to provide prognostic markers capable of predicting whether a cancer is likely to quickly spread into a life-threatening condition.