Seminar: Julia Merta
On January 7, 2026, during our department's seminar, our PhD student Julia Merta, BEng, MSc, gave a captivating presentation entitled "The impact of the batch effect on the effectiveness of MIL models in the classification of patients with endometrial cancer."
Endometrial cancer is the most common cancer of the reproductive organs in women in developed countries. Its diagnosis and treatment depend strictly on the stage of advancement and the histological degree of malignancy of the tumor. With the development of technology, it has become possible to digitize entire microscopic preparations using WSI (Whole Slide Imaging) scanners. This has paved the way for the use of advanced machine learning methods in histopathological analysis.
During her presentation, Julia discussed the important issue of the so-called “batch effect” – technical differences resulting from different staining or scanning protocols that can distort the reliability of AI models. A method for reducing these differences was presented and its impact on the classification of patients according to tumor grade using multiple instance learning models was discussed.
