Point-of-care Colorimetric Analysis through Smartphone Video.

Point-of-care (POC) tests often rely on smartphone image methods for colorimetric analysis, but the results of such methods are frequently difficult to reproduce or standardize. The problem is aggravated by unpredictable image capture conditions, which pose a significant challenge when low limits of detection (LOD) are needed. Application-specific smartphone attachments are often used to standardize imaging conditions, but there has recently been an interest in equipment-free POC colorimetric analysis. Improved output metrics and preprocessing methods have been developed, but equipment-free imaging often has a high LOD and is inappropriate for quantitative tasks. Additional work is necessary to replace external smartphone attachments with algorithms. Towards this end, we have developed a video processing method that synthesizes many images into a single output metric. We use image features to select the best inputs from a large set of video frames and demonstrate that the resulting output values have a stronger correlation with laboratory methods and a lower standard error. The developed algorithm only requires 20 seconds of video and can easily be integrated with existing processing methods. We apply our algorithm to the NS1-based sandwich ELISA for Zika detection and show that the LOD is two times lower when our video-based method is used.