| Abstrakt: |
A Convolutional Neural Network (CNN) model was developed to accurately and efficiently detect inhibition zones in disk diffusion tests for assessing antibiotic resistance and efficacy. Training data was automatically generated considering various experimental conditions, and data augmentation was performed on disk images and inhibition zones to prevent overfitting. The developed model takes 3-channel RGB images as input and outputs 1-channel grayscale images, achieving a final accuracy of 98.17%. By applying the Hough Circle Detection algorithm to the model's output, the position of disks in the culture medium could be precisely detected, and the size of the inhibition zone could be determined by analyzing color changes around the detected disks. The model developed in this study is expected to significantly improve the efficiency and accuracy of antibiotic resistance and efficacy experiments by automating the analysis of disk diffusion test results, which was previously done manually. This can be usefully applied in the fields of antibiotic resistance research and new drug development. [ABSTRACT FROM AUTHOR] |
