Pancreatic cancer Diagnosis with Evolutionary Algorithm

One of the major origins of deaths resulting from cancer in industrialized societies is Pancreatic Cancer [1]. It is also the type of organ that has the least favorable level of prognosis among different types of cancer. Located behind one’s stomach, Pancreas is a large organ shaped like a fish or a triangle. Although it plays an important role in one’s digestive system, we often ignore its importance up until the time something is wrong with it. The main responsibility of Pancreas is to secrete some of the important enzymes that help our body absorb foods, particularly the fat. It also generates insulin and hormones, which are in charge of controlling our body’s blood sugar level. The potential patient’s health can be seriously affected once the balance of these said enzymes or hormones is disturbed. It is essential to have a healthy pancreas if one is looking for a healthy body and a long life. The present study focuses on finding ways for early detection of Pancreatic Cancer through pinpointing a minimal set of genetic biomarkers. These biomarkers will be use to establish diagnosis through the help of the evolutionary algorithm or EA. During the last decade, developing the Evolutionary Algorithm or EA stands among the most important sets of search and optimization methods. EA can be define as modern met heuristics, being use in many different applications, which are categorized as complex. A completely new field known as Evolutionary Computation or EC, has been established based on the success on solving difficult problems. The flexibility gains and their related fitness towards the objective targets in combination with a vigorous behavior are among the major benefits of using EC. During recent years, the application of EC is being considered as a concept that can be adapted to resolve problems, specifically, those with the complex optimization nature.We compare different kinds of Evolutionary algorithm to each other to find the best way for classifying and selecting attributes. We test it on data set, which contains the information from Patient who suffered from this cancer. Our results show that, using MLP for classifying and PSO (particle swarm optimization) for selecting attributes is the best option, which can give us a better result.

کلیدواژه ها:

pancreatic cancer ، multilayer perceptron ، particle swarm optimization ، Genetic algorithm ، Evolutionary algorithm