Media of University of Babylon | Software Department Faculty Member Publishes Research on Integrated Prediction of Image-Based Cell Phenotypes Using Smart Networks

Media of University of Babylon - كلية الطب

Software Department Faculty Member Publishes Research on Integrated Prediction of Image-Based Cell Phenotypes Using Smart Networks

by : Duhaa Fadill Abbas

Date : 20/12/2025 54:18:04 PM

Veiw : 91

Software Department Faculty Member Publishes Research on Integrated Prediction of Image-Based Cell Phenotypes Using Smart Networks

Duhaa Fadill Abbas
In line with its commitment to advancing scientific research and applying artificial intelligence in biomedical fields, Assistant Professor Dr. Sura Zaki Naji, Head of the Software Department at the College of Information Technology – University of Babylon, in collaboration with Master’s student Tayba Hussein Shaman, has published a distinguished research paper in the proceedings of the IEEE International Conference on Internet of Things and Intelligence Systems (IoTaIS 2025). The conference, organized by the IEEE Communications Society (IEEE ComSoc) Indonesia Chapter, is recognized as a highly reputable peer-reviewed international forum.
The study, titled “Integrated Prediction of Image-Based Cell Phenotypes Using Masked Siamese Networks (MSN)”, addresses one of the main challenges in analyzing high-content fluorescence microscopy images used for cellular phenotype characterization. Traditional methods often rely on extensive manual annotation or single-cell segmentation, which is costly, prone to errors, and may fail to capture the true complexity of cellular phenotypes.
The researchers proposed an advanced self-supervised framework based on Masked Siamese Networks (MSN) and built on a Vision Transformer architecture. This approach compares representations extracted from masked views of the same image, allowing the model to identify essential cellular components without requiring prior annotations or segmentation. The system demonstrated high efficiency when applied to the BBBC021 dataset, achieving 97% accuracy in the not-same-compound (NSC) setup and 95% accuracy in the not-same-compound-and-batch (NSCB) setup. Attention map evaluations further confirmed that the extracted representations are structurally sound and biologically meaningful.
The paper was published in the proceedings of IoTaIS 2025, where all accepted papers are indexed in IEEE Xplore and Scopus, ensuring broad scientific dissemination and international academic impact.
This research represents a significant contribution to applied artificial intelligence and bioinformatics, reflecting the College of Information Technology at the University of Babylon’s dedication to supporting rigorous research and fostering innovative academic outputs that keep pace with global advancements.

Research Link :
https://ieeexplore.ieee.org/document/11282097