An automated lateral flow assay identification framework: Exploring the challenges of a wearable lateral flow assay in mobile application

Wearable lateral flow assay (LFA) devices in the point-of-care (POC) diagnosis are an emerging technology due to their ease of use, affordability, and robustness. Recent advances in wearable devices allow rapid self-diagnosis using colorimetry. Visual readouts of the LFA strips often lead to an inaccurate readout due to different lighting conditions. Smartphone-based readouts are increasingly popular among POC platforms. Myriad challenges exist in smartphone-based colorimetric analysis. Image processing on wearable devices faces challenges such as environment (indoor or outdoor), reflection (camera flash, shadow, etc.), different lighting conditions, illumination intensity (low or high), blurred, and a bent wearable sensor on the skin. In this research, an automated framework that tackles these challenges through a series of image processing steps is proposed. A preliminary study that focuses on tackling these challenges through some preprocessing steps and object detection is explored. Our framework accurately identifies the wearable sensor on the body under various conditions. In addition, under some challenging conditions, the biomarker lines are identified in the LFA. A strategy to handle reflection, and high or low illuminations due to camera flash on the wearable LFA is illustrated that improves the sensitivity of the proposed framework. As opposed to the previous frameworks that process the image under a white background, our study captured the real-time images of LFA and used them for the analysis. This research proposes image processing steps for a prospective mobile application in LFA analysis.

Automated summary

This study proposes an automated framework for addressing the challenges of wearable lateral flow assay (LFA) devices on smartphones through image processing steps. The framework accurately identifies wearable sensors under various conditions and detects biomarker lines in the LFA. A strategy to handle reflection and different illumination conditions improves the sensitivity of the framework, which uses real-time images for analysis.