Detection and Identification of Pancreatic Cancer Using Probabilistic Neural Network

The fact that pancreatic cancer has a low life expectancy, that is only 9% of people survive five years, makes a diagnosis catastrophic. The majority of patients are diagnosed late in life, where care choices are minimal. Early diagnosis of pancreatic cancer will greatly increase a person’s chances of survival. Accurate PC staging will help doctors have the right treatment plan for PC patients at different stages, as well as the diagnostic measures needed for a quicker cancer recovery. In this proposed project, ultrasound images will be analyzed. The noise in the image is minimised using the Median Filter. In the next step, Gray Level Co-occurrence Matrix (GLCM) and Discrete Wavelet Transform (DWT)are used to extract related features. Following this extraction step, the refined characteristics are fed into a Probabilistic Neural Network (PNN) neural network classifier, which determines whether or not cancer is present. Metrics such as sensitivity, precision, and specificity are used in experimental computation.