Designing A Machine Learning And Deep Learning Frameworks For Early Pancreatic Disease Diagnosis

Abstract

Cancer is a worrying disease which emerge for detection of medical health. Early detection plays a vital role in physicians providing suitable treatment and increasing the chances of patient survival. When pancreatic cancer is discovered, cancer is well-developed. With the late and incurable point of diagnosis and noteworthy chemo-resistance in tumors, poor results are obtained. Most of the treatments are hopeless in later diagnosis. Many findings reveal the benefits of the premature stage of pancreatic discovery. In early detection, biomarkers are the noteworthy ones that analysis the people in high-risk groups and prioritize them for screening. Thousands of researches accomplished but, no one biomarker for the early detection of cancer converted into a medical application. Obtaining adequate samples for biomarker expansion is difficult during the diagnosis since it needs extensive national and international collaborations. Pancreatic tumors are tremendously diverse within and between people. Thus, accurate diagnosis is the key issue in conventional detection mechanisms. This leads to degrading the overall performance of disease diagnosis. To address these issues, novel machine learning as well as deep learning techniques are developed using proposed research work. The first work aims at segmenting and classifying the MRI/CT images for pancreatic disease using a deep learning-based method. The second and third works aim at enhancing the classifier performance in machine learning to diagnose pancreatic cancer depending on the extracted medical data.