Telight develops advanced microscopes for scientific applications. Their technologies are used in laboratories, research centres, and universities worldwide. One of the company's key products is the Q-Phase microscope, which uses Quantitative Phase Imaging technology to enable scientists to observe cells in their natural environment.
To further enhance the quality and speed of image data processing, Telight decided to add automated cell segmentation using artificial intelligence. In collaboration with the EDIH Ostrava and utilising IT4Innovations supercomputers, particularly Karolina, a customised algorithm for image normalisation was developed, and several deep-learning architectures for cell segmentation were tested.
This collaboration has significantly accelerated and improved cell analysis. Segmentation accuracy has increased by up to 58 %, and the entire process now functions as a true one-click solution, allowing scientists to devote more time to their research.
"Thanks to this collaboration, our software now includes advanced artificial intelligence models. Cell segmentation is faster, more reliable, and up to 58% more accurate," commented Jan Hazucha, Head of Research and Development at Telight, on the collaboration.
"We trained hundreds of artificial intelligence models on tens of thousands of images on the Karolina supercomputer. This meant that we didn't miss any high-quality models and at the same time managed to significantly reduce the time needed to find them," said Vladimír Ulman, Senior Researcher at IT4Innovations.
Watch the videospot about the collaboration with the Telight company.
* The collaboration was realized thanks to the European Digital Innovation Hub Ostrava project.