The Applied and Research Center for Pharmaceutical Drugs and Technologies at Karadeniz Technical University (KTÜ İLAFAR) has developed AI-powered software to address a common problem in microscopy: part of a tissue or cell appears clear while other parts remain out of focus. The project aims to produce clear images across the sample, rather than limiting clarity to a single point within the field of view.
The software was developed as part of the project “Development of New Deep-Learning Methods for Creating Optimally Focused Three-Dimensional Images and Shapes in Microscopic Systems,” carried out by KTÜ İLAFAR in cooperation with Trabzon University and Ohio State University. The center’s director, Professor Feride Sena Sezen, said work on the project began about four years ago and that the team put the software into use after developing it.
What changes in practice?
Doç. Dr. Hülya Doğan, head of the Innovative Artificial Intelligence Applications Group at KTÜ İLAFAR, explains that the issue is not limited to the clarity of a single image but also concerns researchers’ ability to read the entire sample. Specialists rely heavily on two-dimensional images to examine cells or assess disease-related indicators, but differences in depth can obscure features that do not appear in an image focused at a single level.
The system adds a three-dimensional dimension to the imaging process, with the aim of providing more detailed information about cells and samples. Doğan also says the software does not merely apply algorithms to enhance unclear areas, because this type of processing may alter the original data; instead, it aims to provide images that contain the sample’s original values.
Applications beyond pharmaceutical research
The project’s initial use is linked to preclinical pharmaceutical research and the processing of data obtained from laboratory animals and cells, in addition to medical and research fields that rely on microscopes. According to the team, the software can also be used in geology and geophysics, chemistry, aquaculture, agriculture, genetics, the food industry, environmental science, veterinary medicine, forestry, and mining engineering.
Doğan notes that the tool was designed to be compatible with different types of samples, microscopes, and cameras, which could broaden its use compared with a solution tied to a single imaging platform.
Why does this news matter?
The project’s core technical value is not the addition of a conventional autofocus feature, but the attempt to combine full-field clarity with preservation of the sample’s data, while converting two-dimensional images into a three-dimensional representation that can provide researchers with additional characteristics. This is important in environments that rely on microscopic images to make research decisions or understand changes in cells and tissues.
However, the source provides no figures on the algorithm’s accuracy, the size of the dataset, or the types of microscopes and samples on which it was tested, nor does it mention results from an independent comparison with available systems. The team says a paper on the method and dataset has been submitted to an international journal and that it is prepared to share the algorithm, software, and data with private-sector representatives. Therefore, the tool’s readiness for broad clinical or commercial use still requires additional verification.