Development of an Automatic Antibiogram Reader System Using Circular Hough Transform and Radial Profile Analysis

The disk diffusion method is one of the most frequently used techniques to determine the antibiotic susceptibility profiles of pathogens. A fast identification of resistance profiles is essential for researchers and physicians. In laboratories without technological resources, these measurements are done manually using a ruler or a caliper increasing the chances of error. We have designed a device that, by mean of image processing of disk diffusion tests, semi automatizes the process. A total of 53 images of plates containing, on average 11 inhibition zones each were acquired and we compared the results obtained by the proposed algorithm with the results from the commercial equipment Osiris^® (considering the correction in the measurements made by a human operator). The proposed algorithm correctly identified the position of the 597 inhibition zones in 100% of the images. The Pearson correlation coefficient between the measurements of the inhibition zones diameters by the proposed algorithm and our reference was 0.7728. The Cohen’s kappa coefficient between reference and the proposed method was 0.729, and the concordance between methods was 86.3%, where the most significant error rate was 0.2%. This study was approved by the HCPA Health Ethics Committee under number 170587.