An Optimized Feature Based Framework for Pancreatic Cancer Detection

Abstract

Pancreatic cancer is a deadliest form of cancer and detecting it at an early stage is challenging. Survival rate for Pancreatic Cancer is low when compared to other cancers like breast, lung, and colon etc. Pancreatic Cancer is expected to become the second leading cause of cancer death worldwide. Time becomes an important factor in identifying the abnormalities. Early diagnosis is essential for proper clinical decision and for surgery. Recent advancement in diagnostic Technology such as CT, endoscopy ultrasound and monitoring the metabolic response remains challenging for this disease. PET/CT a relatively novel modality, and widely used in oncology and achieves best results in pancreatic cancer. Pancreatic cancer is detected and identified using automated machine learning techniques. New Model for cancer detection was developed based on PET/CT images. Computer Aided Diagnosis has become an encouraging tool for helping radiologists and physicians in identifying the cancer accurately.The major problem is the difficulty in identifying the pancreatic cancer and the survival rate is very less because the cancer cannot be identified at an early stage. Early diagnosis of the disease and proper treatment can save life. PET/CT scan images can determine the extent of metastasis. Usage of PET/CT imaging is improving in medical field to diagnose the cancer at an early stage. Several systems have been developed and work is under way to detect pancreatic cancer using PET / CT imaging technique, although some existing systems have given unsatisfactory results. Hence, to overcome these problems, an optimized feature based framework has been proposed to detect and classify the pancreatic cancer. Computer Aided Diagnostic system have become an essential tool for diagnosing the pancreatic cancer at an early stage. This work proposed a novel methodology and creates a model for the early diagnosis and classification of pancreatic cancer using medical modality of PET/CT