Application Of Deep Learning For Gastrointestinal Disease Detection Using Endoscopic Images

Abstract

The scarcity of trained professionals is one of the major causes of apprehension as far as newlinethe healthcare system in India is concerned. Factors such as unhealthy or unbalanced food habits, newlineerratic work culture, increased stress, and lack of exercise attribute to an increase in gastrointestinal newline(GI) problems in middle and high society. On the other hand, malnutrition in children, and newlineunhygienic environment in slums and rural areas contribute to the proliferation of GI problems newlineamongst poor and underprivileged families in India. It can very well be imagined the predicament newlineof gastroenterologists. One of the significant roles of a gastroenterologist is to visualize and newlineanalyze images and videos of the GI-tract. There happen to be different GI-tract imaging newlinetechniques, most of which involve non-invasive procedures such as X-ray, magnetic resonance newlineimaging, computed tomography, ultrasound, and positron emission tomography. Apart from these, newlineendoscopy, which is a minimally invasive procedure, provides a very detailed and clear image. newlineEndoscopy also facilitates biopsy and treatment of the GI-tract which are not possible with other newlineavailable modalities. Needless to say, an early diagnosis of GI disease reduces the risk of critical newlinedisease conditions, and this requires the skill of experienced and expert professionals. With newlinegrowing cases of GI patients, the data to be analyzed is also proportionately increasing the burden newlineon gastroenterologists. This demands an efficient smart healthcare system, and this requirement newlineforms the prime motivation behind this study to explore the use of artificial intelligence (AI) based newlinetechniques for fast and reliable diagnosis of GI-tract diseases. newlineTo begin with, the study makes an attempt to present a bird s eye view of the role of AI in newlinethe domain of healthcare along with an allusion to the area of gastroenterology and its medical newlineimaging modalities. This is followed by a generalized literature study identifying the technical newlinedifferences between machine learning and deep learning (DL). Th