2020-855 An Automated, Cost-Effective Optical System for Early Antimicrobial Susceptibility Testing Using Deep Learning

SUMMARY

UCLA researchers in the Department of Electrical and Computer Engineering developed a cost-effective optical apparatus that allows for fast and accurate antimicrobial susceptibility test readings by detecting bacterial growth in short incubation times and eliminating human error.

BACKGROUND

Antimicrobial resistance (AMR) is the ability of microorganisms to resist antimicrobial treatment such as antibiotics and it is responsible for 2.8 million cases and 35,000 deaths just in the U.S.A. A common tool to diagnose a patient’s infection with AMR is the performance of an antimicrobial susceptibility test (AST) to determine which antibiotic treatment will be the most effective against a given patient’s bacterial infection. An AST is performed by taking a patient’s bacterial sample and incubating it for 18-24 hrs, divided in a 96-well microplate to test different antibiotics/concentrations in each well and visually inspected by a trained diagnostician. However, the lengthy incubation times put patients at risk because in the interim they may be prescribed strong broad-spectrum antibiotics or may be prescribed an ineffective antibiotic. Furthermore, samples are visually inspected which may introduce human error and variability. Therefore, there is a need for an automated, cost-effective system that can speed the detection and quantification of resistance in AST.

INNOVATION

UCLA researchers in the Department of Electrical and Computer Engineering developed a cost-effective apparatus that delivers early AST results by reducing the incubation time and sample reading as well as eliminating human errors, while compatible with standard clinical workflow. The apparatus, composed of off-the-shelf optics and low-cost Raspberrry Pi computers, was capable of identifying grown bacteria only after 5.72 of incubation in the wells at a 95.03% accuracy. Moreover, this optical system met the FDA-defined criteria for essential and categorical agreement testing 14 drugs. The apparatus enabled faster, more accurate, and less expensive AST, which could accelerate the current technologies in quality control and antimicrobial resistance research.

POTENTIAL APPLICATIONS

• Antimicrobial susceptibility test
• Clinical quality tests
• Antibiotic research
• Diagnostics in agricultural and animal growth

ADVANTAGES

• Low-cost
• Higher accuracy than current methods
• Fast bacterial detection

RELATED MATERIALS

• Calvin Brown, Derek Tseng, Paige M. K. Larkin, Susan Realegeno, Leanne Mortimer, Arjun Subramonian, Dino Di Carlo, Omai B. Garner, and Aydogan Ozcan. Automated, Cost-Effective Optical System for Accelerated Antimicrobial Susceptibility Testing (AST) Using Deep Learning. ACS Photonics. 2019. DOI: 10.1021/acsphotonics.0c00841.
• Tok S, de Haan K, Tseng D, Usanmaz CF, Ceylan Koydemir H, Ozcan A. Early detection of E. coli and total coliform using an automated, colorimetric and fluorometric fiber optics-based device. Lab Chip. 2019;19(17):2925-2935. doi:10.1039/c9lc00652d.

STATE OF DEVELOPMENT

First successful demonstration and reduction to practice.