Development of Novel Method for Detection of Gastric Cancer by Hyperspectral Image Technique

Abstract

Over the last few decades, the gastric incidence all over the world has been newlinechanged dramatically. As per the statistics goes, before 1950 in United States of newlineAmerica, gastric cancer was the predominant cause of cancer death in men and also it newlinewas the third leading cause of cancer death in women. Now due to the changes in newlinedietary, the mortality rate of gastric cancer in has been drastically declined in USA. newlineMedical diagnosis is principally supported by the imaging techniques such as newlineMRI, CT, Ultrasound, Doppler scanning and nuclear imaging have completely newlineexpanded medical imaging field. Modern research works showed that Hyperspectral newlineimaging (HSI), has emerged as a new member of the family of the medical imaging newlinemodalities. The HSI system caters as one of the powerful non-invasive tool for tissue newlineanalysis. This method is able to capture both the spectral and spatial information of an newlineorgan or tissue in one image. In other words, this modality gives narrow band images newlineat different wavelengths. It is not similar to traditional three channel color cameras and newlineother imaging systems which are based on filters. This system capture full neighboring newlinespectral as well as spatial information. This work will be used for detection of gastric newlinecancer by using developed novel algorithm. newlineAnother contribution to this work is in detecting ulcer and blood congestion in newlinetissues. This technique was illustrated by capturing an image of diagnostic patient newlinefollowed by preprocessing in wavelet domain. Once the preprocessing is done, tumour newlineregion is identified we would go further to extract statistical information from the tumor newlinesuch as area of the tumor. When any treatment is carried out after diagnosis the subject newlineor the doctor would be interested in knowing if there is any improvement or if the newlinemedication is giving results as planned. To know that area or the texture would give a newlinebetter insight hence area of the tumor would be extracted using region properties, newline newlineestimation and hence would aid in knowing the shrinkage of the tumor in subsequent newlinediagnosis. The entire visualization of Gastro Intestinal (GI) tract is examined in this newlinework. This work is distributed into 4 modules i.e. Image analysis, treatment newlineperformance, blood detection and classification of the image using Support Vector newlineMachine (SVM). The automatic detection of gastronomical diseases were carried out newlineaccurately without any ambiguity. newline