Currently, I am working on a collaborative project between radiologists from Johns Hopkins University and Center for Visual Computing (Stony Brook University) to determine the most descriptive radiological features and develop a system for pancreatic cyst classification.
Pancreatic cancer is one of the deadliest of all solid malignancies with a 7% five-year survival rate . It is the third leading cause of cancer-related mortality in the United States. There are no specific early symptoms of pancreatic cancer , and most of the cases are diagnosed at a late stage and spread beyond the pancreas. Due to this feature of the disease, less than 20% of patients are candidates for surgery . However, recent research highlighted cysts as the precursors to pancreatic cancer . Hence, it is utmost importance to identify and remove malignant cysts in the early stages of pancreatic cancer to increase the survival chances of a patient.
Today the most common pancreas screening procedure is Computed Tomography (CT). Standard abdominal CT screening has the sensitivity of 70-80% for pancreatic lesions less than 2 cm in diameter . National Cancer Institute set the development of the noninvasive techniques for detection and monitoring of the disease as one of the critical priorities for pancreatic cancer screening. Computer-aided diagnosis (CAD) systems play an important role in this priority. CAD systems can help with the surgery planning and navigation, as well as the automatic detection and the assessment of the disease. CT is a logical input for CAD systems: it is relatively cheap, easily accessible and it has a high-spatial resolution, which provides a large amount of information.
Segmentation of the organ is considered an initial step of the most CAD systems. However, high inter-patient variations in size, location, and a lobulated shape of the pancreas make a manual segmentation challenging even for an experienced radiologist. The visceral fat tissue in the proximity of the organ causes large variations in contrast of the boundaries of the pancreas. Moreover, pancreatic masses distort the appearance of the organ making the detection and segmentation of the organ significantly more difficult.
Classification between different kinds of cysts, malignant and benign masses is a second step of the CAD systems. The variability of the imaging characteristics of the pancreatic cysts  expands the number of challenges in the diagnostic of the disease. Thus, understanding of the different radiological features of the normal and abnormal pancreas and its masses is essential in the development of the CAD system.
Pancreatic cysts can be split into two categories - neoplastic cysts and inflammatory cysts . Inflammatory cysts, or pseudocysts, are the most common pancreatic lesions and are the results of the pancreatitis or trauma. When the diagnosis of the pancreatitis is excluded, a case of a neoplastic cyst must be considered. Classification of the neoplastic cysts is based on the type of the fluid that these cysts produce. There are two major groups of the neoplastic cysts: mucinous and non-mucinous cysts. Mucinous cyst include Mucinous Cystic Neoplasms (MCN) and Intraductal Pappilary Mucinous Neoplasms (IPMN). The later category consists of Serous Cystic Neoplasms (SCN) and other solid lesions, e.g. Solid Pseudopapillary Neoplasms (SPN). It is utmost important to differentiate between them to make the right treatment decision.
 American Cancer Society. "American Cancer Society: Cancer Facts and Figures 2016." (2016).
 Porta, M., Fabregat, X., Malats, N., Guarner, L., Carrato, A., de Miguel, A., Ruiz, L., Jariod, M., Costafreda, S., Coll, S. and Alguacil, J., 2005. Exocrine pancreatic cancer: symptoms at presentation and their relation to tumour site and stage. Clinical and Translational Oncology, 7(5), pp.189-197.
 Hruban, R.H., Maitra, A., Kern, S.E. and Goggins, M., 2007. Precursors to pancreatic cancer. Gastroenterology Clinics of North America , 36(4), pp.831-849.
 Steinberg, W.M., Barkin, J.S., Bradley III, E.L., DiMagno, E., Layer, P., Canto, M.I. and Levy, M.J., 2009. Should patients with a strong family history of pancreatic cancer be screened on a periodic basis for cancer of the pancreas?. Pancreas, 38(5), pp.e137-e150.
 Lennon, A.M. and Wolfgang, C., 2013. Cystic neoplasms of the pancreas. Journal of Gastrointestinal Surgery, 17(4), pp.645-653.