Discipline: Technology and Engineering
Subcategory: Electrical Engineering
Kyle Liang - University of California, Los Angeles
Co-Author(s): Hatice Ceylan Koydemir, Steve Feng, Rohan Nadkarni, Derek Tseng, Parul Benien, and Aydogan Ozcan, UCLA, Los Angeles, CA
Every year, millions of people are affected by Giardia lamblia, a waterborne parasite. This protist causes Giardiasis, responsible for various symptoms including e.g., stomach cramps, nausea, and diarrhea. One of the most effective methods to prevent the spread of this disease is through rapid and on-site detection, which is not always possible in resource-limited settings. Here we present a field-portable, cost-effective, and easy-to-use mobile phone based microscopy device which can automatically detect and count Giardia cysts within water samples. This device utilizes a smartphone based fluorescence microscope, a disposable sample processing cassette, and a custom-developed application. This handheld microscope design has a large field-of-view of >0.7 cm2, weighs only ~180g, and operates with the use of two AA batteries and a smartphone. The sample processing cassette is capable of large volume storage as it consists of a porous filter membrane which has a pore size of 5 microns to prevent backflow and to capture Giardia cysts, and cotton absorbent pads that can hold up to ~20 mL of water sample using a flow through assay. Our custom-built application provides an easy interface for any untrained technician to operate our handheld microscope by taking an image, which in turn is sent to our servers (local or remote) for automated detection and counting of Giardia cysts captured on the membrane using our image processing algorithms and machine learning data, which consist of >30,000 cysts with labels and 96 features per cyst. The cyst counting results are sent back to the user through the same application, removing the need for a trained expert to detect Giardia cysts. This entire analysis takes less than an hour for a 10mL water sample. We compared the sensitivity and the specificity of our platform using various supervised classification models, including support vector machines (SVMs) and nearest neighbors (NN), and demonstrated that a bootstrap aggregating or bagging approach provides the highest classification accuracy. We also evaluated the performance of this machine learning based mobile platform with different sources of water, e.g., tap water, non-portable water and pond water, and achieved a limit of detection of 12 cysts per 10 mL, an average cyst capture efficiency of ~79%, and an accuracy of ~94%. This field portable and cost-effective platform could be useful for water quality monitoring in resource-limited settings.
Funder Acknowledgement(s): The Ozcan Research Group at UCLA gratefully acknowledges the support of the Presidential Early Career Award for Scientists and Engineers, the Army Research Office (ARO); W911NF-13-1-0419 and W911NF-13-1-0197), the ARO Life Sciences Division, the National Science Foundation (NSF) CBET Division Biophotonics Program, the NSF Emerging Frontiers in Research and Innovation (EFRI) Award, the NSF EAGER Award, NSF INSPIRE Award, NSF Partnerships for Innovation: Building Innovation Capacity (PFI:BIC) Program, Office of Naval Research, the National Institutes of Health, the Howard Hughes Medical Institute, Vodafone Americas Foundation, the Mary Kay Foundation, Steven & Alexandra Cohen Foundation, and KAUST. This work is based upon research performed in a laboratory renovated by the NSF under Grant No. 0963183, which is an award funded under the American Recovery and Reinvestment Act of 2009.
Faculty Advisor: Aydogan Ozcan, ozcan@ucla.edu
Role: One of the co-authors wanted a way to both confirm which objects the machine learning algorithm incorrectly labelled and to see which images produced the best results for the machine learning algorithm. My job was to create an interface implementing image processing algorithms to map images of Giardia from different cameras to one base image that was labelled with Giardia to confirm if the data the machine learning algorithm had labelled were valid Giardia cysts.