Advanced search
Start date

Technology for automating the process antibiogram analysis based on disk diffusion

Grant number: 20/09863-5
Support type:Research Grants - Innovative Research in Small Business - PIPE
Duration: May 01, 2021 - January 31, 2022
Field of knowledge:Biological Sciences - Microbiology - Applied Microbiology
Principal researcher:Feng Chan Jung
Grantee:Feng Chan Jung
Company:Feng Chan Jung Tecnologia da Informação Ltda
CNAE: Desenvolvimento e licenciamento de programas de computador customizáveis
Atividades de serviços de complementação diagnóstica e terapêutica
Atividades de apoio à gestão de saúde
City: São Paulo
Assoc. researchers:Bruna Trentinaro Ibiapina


The increase in bacterial resistance to antibiotics and the appearance of well-known super bacteria reveal a scenario of extreme relevance globally, generating a great effort by authorities and institutions to monitor and control this situation in Brazil and Worldwide. The most widely method used to test the sensitivity or resistance of a bacteria to an antibiotic is an in vitro technique called disk-diffusion antibiogram. However, this method is pure manual and it highly depends on the experience of the biologist, also it time consuming (such delays in generating results), overload of the biologist, exposure to error on manual transcription causing variations in the results. The main objective of this PIPE Project is to develop and validate a local technology within low cost and high performance for antibiogram testing using disk-diffusion method aiming to automate the process and its outcomes by incorporating Artificial Intelligence (AI) algorithms, democratize the technology as an alternative to looking for low cost, generating agility/productivity, standardization, and quality of results. The research comprises the development of the algorithm for image acquisition of the disk-diffusion plate, its interpretation according to the current BrCast Standard, creation of the ad-hoc report and the real-time validation of the results. The development of the algorithm for obtaining the image of the disk-diffusion plate will be done using the latest generation of programing language with advanced technology for image interpretation, composed by Python ( currently Python is the most used language today to create and train vision models computational. The validation of the results will be made in clinical samples by comparison between results obtained manually and the results obtained by ASTeK ( to be further analyzed statistically). The methods for automating the disk-diffusion antibiogram process have four main challenges: identification of the disks and their positions, identification of the alphanumeric labels of each disk, identification of the presence or absence of a halo and calculation of the diameter of the halo if any. The main uncertainty is whether the image acquisition system and the application of the algorithm will provide safe and robust results, especially in relation to non-traditional images, which is considered very challenging. The solution proposed here has the expected result of overcoming these challenges and this solution includes software development, the intense use of IA such Computer Vision, Machine Learning and Deep Learning. As a result of this phase of the project, it is expected to obtain the development of automated ASTeK for the interpretation of the antibiogram exam and its respective technical validation through the results collected and final tuning of the level of adherence of the algorithm developed. One of the steps recommended in the final report generated by the "Review on Antimicrobial Resistance" is the promotion of rapid and effective diagnostic tests, aiming at not using antibiotics unnecessarily, and the ASTeK project is in line with this recommendation. According to a projection of this same study, in 2050 bacterial resistance to antibiotics will be responsible for the death of 10 million people, becoming the main cause of death in the world, with costs for the global economy of the order of 100 trillion dollars, if not no action is taken with a view to solving this problem. Another recommendation in this report is to increase global surveillance of antibiotic resistance. In this sense, the proponents of this PIPE Phase I project (ASTeK) are developing, with their own resources, an informational intelligence system (Big Data) that, together with the data from ASTeK, will enable the offering of a technological informational surveillance platform, useful private and public health for clinical laboratories and public health agents in preventive and / or corrective actions. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items