ALIMUL KHAN 2018 DESIGN OF AN OPTICAL SENSOR TO DETECT HUMAN INTESTINAL BLEEDING FOR CAPSULE ENDOSCOPY

Abstract

Gastrointestinal (GI) bleeding in human is not uncommon that may sometime lead to fatal consequences. Types of GI bleeding include acute and chronic bleeding, upper and lower bleeding. In some cases, the patient requires immediate diagnosis. These abnormalities could be detected by using a flexible wired endoscope, a tool to see inside view of the GI tract. An advanced version of endoscopy tool is the wireless capsule endoscopy (WCE), first used in 2001, where the patient swallows an electronic capsule-shaped device which captures thousands of images and sends to an external data recorder via wireless communication. The images are later analyzed by physicians to detect GI abnormalities. The screening process is time-consuming. Delay in bleeding detection may increase the risk of the patient. Therefore, an easy and faster way to detect bleeding in the GI tract by using optical sensors is warranted. Standard methods to diagnose GI bleeding are still limited due to unpredictable bleeding behaviours, such as the rate of bleeding and the concentration of blood. In addition, proper instrumentation, lack of easy access and control of the sensors are also limiting the techniques of bleeding detection. Blood can be identified by analyzing its optical properties. To determine its optical properties, three types of experiments using a spectrophotometer, a pulse oximeter, and an RGB colour sensor were conducted in this research. The principle is to emit light of a specific spectrum from a source on a liquid sample. Depending on the protein (like hemoglobin) or other substances present in the liquid, a different wavelength of light will be reflected and/or transmitted, which is then detected using a sensor (typically the photodiode). The objective of this study is to distinguish blood samples (BS) from non-blood samples (NBS) using their optical properties. Different concentrations of horse blood, swine blood, and bovine hemoglobin solutions were used in this experiment as BS. Many other solutions like coffee, tea, juice, coloured water, etc. were used as NBS. One set of data samples was first used to develop the algorithm. Next, another set of samples was used to validate it. Several optimum light spectrum wavelength were identified. Analysis of the experimental results suggests that it is possible to separate BS from NBS using the optical properties of the substances. The outcomes of this study can be used to implement small optical sensors in a capsule endoscopy system to detect GI bleeding. The optical sensor will eliminate the post-processing of the images and thus save a significant amount of time.