Master of Science (MS)
computer vision, feature extraction, image processing, kidney disease monitor, machine learning, smart phone application
The number of people diagnosed with advanced stages of kidney disease has been rising every year. Early detection and constant monitoring are the only way to prevent severe kidney damage or kidney failure. Current test procedures require expensive consumables or several visits to the doctor, which results in many people foregoing regular testing. To address this problem, we propose a cost-effective teststrip-based testing system that can facilitate kidney health checks from the comfort of one’s home by using mobile phones. The specially designed teststrip facilitates a colorimetric reaction between alkaline picric acid and creatinine in a blood sample that has been applied to the teststrip. Our system uses state-of-the-art deep learning localization models to capture quality images of the teststrip using a cell phone, and then processes them using computer vision and machine learning techniques to predict the concentration of creatinine in the sample based on the change in color. The predicted creatinine concentration is then used to classify the severity of the kidney disease as normal, intermediate risk, or kidney failure. We thoroughly evaluate the effectiveness of our models, both in the localization and classification tasks, and find that our histogram of color-based, hybrid nearest neighbor methods outperform alternatives and exhibit good overall prediction performance.
Ramesh, Rathna, "Machine Learning Methods for Kidney Disease Screening" (2018). Master's Theses. 4951.
Available for download on Wednesday, October 14, 2020