The main task of preclinical studies in toxicology and pharmacology is to determine the toxic properties and pharmacological effect of the active ingredient. In these studies, researchers often use freshwater fish. Due to the biological similarity of about 70% of genes with human genes, speed and convenience of work in laboratories, scientists use Danio rerio, a small freshwater tropical fish.
Maxim Nigmatulin, a 3rd-year student of the Faculty of Computer Science and Technology of ETU “LETI”, aims to make the analysis of new drugs easier with his project titled “Development of Software for Analyzing the Behavior of Danio rerio using Computer Vision.”
The young researcher is developing the project with the help of a grant from the UMNIK Foundation for Assistance to Innovations under the supervision of Dmitry Kaplun, Assistant Professor of the Department of Automation and Control Processes of ETU “LETI.”
The development is based on computer vision and artificial intelligence technologies. The software recognizes the fish on the downloaded video and then analyzes the parameters of its movement: if the fish moves quickly, the drug has a stimulating effect; if slow, then sedative.
“The proposed approach will reduce the time spent on analyzing the behaviour of Danio rerio and the impact of human factor on the results of the experiment, creating an affordable alternative to more expensive foreign counterparts. The developed solution will also be extensible: it will allow considering new parameters of the experiment, which are difficult to estimate with visual observation,” Maxim says.
Compared to existing analogues, the new software will be much cheaper and more mobile. It will allow laboratories to conduct research faster and more profitable.
“The software will run on various operating systems such as Windows, Linux, and macOS. Ultimately, the project will provide a lightweight, affordable, multiplatform, extensible user application,” Maxim explains.
Now, the project is in the first stage of development. Maxim is creating an algorithm for video data preprocessing in Python, an algorithm for final functionality in 2D and 3D, and a module library based on the developed algorithms.
At the second stage, the young researcher plans to develop a visual interface in PyQt, study docker technology and build a docker container for the application to make it multiplatform, and release an alpha version for testing.
The result will be complete software that can be used on different operating systems to quickly and efficiently analyze Danio rerio behaviour patterns.