Pre-Writing Seminar for Master’s Student Tiba Hussein Shamman Hussein in the Department of Software

Duhaa Fadill Abbas

As part of the ongoing scientific activities in the Department of Software at the College of Information Technology – University of Babylon, Master’s student Tiba Hussein Shamman Hussein presented her pre-writing seminar for her thesis titled: “Image-Based Prediction of Cellular Phenotypes,” under the supervision of Asst. Prof. Dr. Sura Zaki Naji, and in the presence of the department’s scientific committee.

In her presentation, the student highlighted the critical importance of understanding cellular phenotypes from microscopic images, emphasizing their vital role in drug discovery and biological research. The study aims to develop a weakly supervised framework for analyzing high-content fluorescence microscopy images of breast cancer cells treated with various chemical compounds at multiple concentrations.

The research employed Masked Siamese Networks (MSNs) to learn stable and biologically meaningful representations of cellular features, guided by weak labels derived from compound treatments, thereby eliminating the need for manual annotations. These learned representations were further integrated with learnable prototypes to enhance consistency and robustness.

Additionally, a post-processing step was applied to the extracted features to reduce technical variability and improve overall robustness. This resulted in stable, comparable embeddings for each treatment, which were subsequently utilized in downstream tasks such as phenotype prediction, classification, clustering, and visualization.