Implementation of image processing for red and white blood cell detection and counting in urine samples: A thesis

• Main Authors: Sanchez, Jan Mason S., Santisteban, Mariel Lian M., Suarez, Kenneth A.
• Format: text
• Language: English
• Published: Animo Repository 2014
• Subjects: Urine > Analysis; Urine > Examination; Blood cells; Blood > Examination; Blood cell count
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/8596
• Institution: De La Salle University

Urinal determines the criteria for diseases that infect the urinary tract and other areas which has the byproduct of urine. Urinalysis remains a potent screening tool for clinicians in this case, we pertain to Microscopic Urinalysis (Williams, 2013). Since there already exists applicants available for public use in Android and iTunes market which provides chemical analysis of urine through the means of image processing and usage of reagent strips, the researchers had undergone the development of microscopic analysis of urine, a more complex urine examination, which aims to conveniently process microscopic urine sample images by counting the number of red and white blood cells and storing the respondents information and urine cell count results in a database, MySQL, that is accessible online to a medical facility. The software prototype utilizes the concept of image processing by tracking contours and shape detection to identify the red and white blood cells which are circular in nature. Upon the creation of the software, the researchers were teamed up with licensed Medical Technologists (RMT) to verify the effectively and efficiency of the devised software prototype. The software prototype consists of a user-friendly graphical user interface (GUI) which serves as the base container of the aimed functionality completed using an open source integrated development environment (IDE), Microsoft Visual Studio, the image processing section applies a library of programming functions which is mainly aimed at real-time computer vision Open Source Computer Vision (OpenCV) and the data and results are being collated in an open source database, MySQL. The researchers based the calibration of the software prototype image processing function parameters on a training set composed of fifteen (15) unique urine microscopic images while a testing set containing thirty (30) samples is then used for the final testing of the devised software prototype. Upon the proposal of the study, a difference margin of 20% - 30% cell count between the licensed MT count and the result of the software prototype was set and is then successfully achieved. The researchers concluded that the study was a success, and the devised software prototype exported as an executable file was able to identify the cells accordingly by being able to detect and count the red and white blood cell concentration in a high power field and store the acquired information and results in database ready for cloud sharing to a medical facility which can further verify and give proper diagnosis in determining early signs of urinary tract infection.