Low-Cost Automated Paper Strip Reader (APSR)

Paper strip analyzer or a medical strip reader is a low-cost Paper-based Analytical Device (PAD) that is easy-to-use and widely accepted Point-Of-Care (POC) alternative to the relatively complex and expensive ELISA (Enzyme-linked ImmunoSorbent Assay) tests. These devices are used for determining the chemical concentrations (pH, glucose, protein, nitrite, etc.) of the body fluids (blood, urine, sweat, saliva, etc.) to monitor organ functionality and ensure early diagnosis of potential diseases. The test paper strips can be analyzed by comparing the colour changes on the test strip to the existing colour chart either by the operator’s visual interpretation or by a smartphone-based automated system which is a fairly new technology from the past decade. Apart from being time consuming, the manual (naked eye) interpretation of results is subject to the robustness of the clinicians’ eyesight and ambient lighting conditions that may potentially introduce an error while obtaining results. This presented the need to automate this process of colour detection. Despite the faster computation speed of smartphone-based systems, they may introduce image acquisition related errors owing to the varying camera resolutions in different smartphone models. Smartphones also face the issue of poor battery backup and software (in)compatibility. There exists a research gap in terms of an automated technology that is ASSURED: A = Affordable, S = Sensitive, S =Specific, U = User-friendly, R = Rapid & Robust, E = Equipment-free, D = Deliverable. These are the requirements proposed by the World Health Organization (WHO) for commercial medical sensors. In this study, we try to overcome the shortcomings of smartphone-based technology while adhering to the ASSURED criteria. We present a fully functioning prototype of a 3D printed Raspberry-Pi microcontroller based paper strip reader with a rechargeable battery and consistent illumination. The device is paired with a backend Python program and a user-friendly GUI to display the results. The device also has a feature to export the results with one click to your clinician via email. APSR algorithm is 88.7% accurate and the speed of operation is≈5s which is faster than the best speed reported in the literature yet. The future scope involves making use of Machine Learning and Deep Learning to further improve the algorithm.Biomedical Engineering | Bioelectronic