همایش ملی بیوانفورماتیک ایران

صفحه اصلی / 4th international edition and 13th Iranian Conference on Bioinformatics

A Novel Approach to Antimicrobial Susceptibility Testing: Automated Disk Diffusion Analysis with Smartphone Cameras and Deep Learning Techniques

نویسندگان :

Mahdiar Mansouri¹ Mohammadreza Najafi Disfani² Danial Ghofrani³ Arash Pournaji⁴

1- Faculty of Life Sciences and Biotechnology, Shahid Beheshti University, Tehran, Iran 2- Faculty of Life Sciences and Biotechnology, Shahid Beheshti University, Tehran, Iran 3- Faculty of Life Sciences and Biotechnology, Shahid Beheshti University, Tehran, Iran 4- Faculty of Life Sciences and Biotechnology, Shahid Beheshti University, Tehran, Iran

کلمات کلیدی :

deep learning،image processing،disk diffusion

چکیده :

Introduction: Antibiotic diffusion tests are essential in clinical microbiology for assessing microbial resistance and antibiotic efficacy. These tests measure inhibition zones around antibiotic disks to determine microbial susceptibility. Traditional methods can be slow and error-prone, particularly when processing large sample sets, underscoring the need for a fast, automated approach. Method: We developed a high-speed automated analysis system using OpenCV and deep learning techniques to detect and analyze antibiotic diffusion disks on plate images. The system utilizes OpenCV’s contour detection for inhibition zone identification and measurement, Canny and Sobel filters for edge detection and image quality enhancement, and UNet for disk segmentation and OCR to read antibiotic labels. Results: The method achieved a 98.7% accuracy rate in detecting and labeling antibiotic disks, accurately identified inhibition zones, and provided precise measurements. Cross-referencing with Clinical and Laboratory Standards Institute (CLSI) guidelines confirmed its effectiveness in determining antibiotic resistance or susceptibility. The system demonstrated rapid processing speeds, maintaining high accuracy even with large sample volumes. Discussion: This approach provides a scalable, efficient solution for antibiotic diffusion test analysis, minimizing human error and supporting high-throughput needs in clinical and research settings. Future work will enhance adaptability to different disk formats and antibiotic types, expanding its potential for broader testing applications.