How to Use Deep Learning for Automatic Segmentation of Brain Tumors – 8 December

During the seminar, Maciej Kuchciak from the Faculty of Economic Sciences will discuss how the use of deep learning for automatic segmentation of brain tumors based on magnetic resonance images and limited computational resources (the full abstract is provided below).

The lecture, titled „Glioma Segmentation Throughout the Treatment Continuum: A Systematic Evaluation of Deep Learning Models for the BraTS 2025 Lighthouse Challenge,” will be held as part of the QFRG and DSLab seminar series.

All those interested in the topic are warmly invited to attend at 16:50 (the meeting will begin at 17:00) in room B002 or to join the event online. To receive the link for online participation, please contact us at: ]7iWc{frm[hqU.zj_u-TP*]#[N*YES_QT*SWW^~rsEZ}]Aq.

After the presentation which will last approximately 45 minutes, we encourage you to take part in the discussion.

---

This research investigates automated segmentation of glioma brain tumors in multi-parametric MRI across pre- and post- treatment stages, focusing on clinically realistic, resource constrained settings with a single Tesla T4 GPU. Using the BraTS 2025 Lighthouse dataset (2,872 cases with four MRI sequences and four annotated subregions), we conduct detailed exploratory analysis of class imbalance, volumetric distributions and morphological differences between pre- and post- operative scans, which motivates a full volume crop-and-pad preprocessing pipeline, intensity normalization and label cleaning. Two deep learning architectures, 3D U-Net and SegResNet, are implemented and compared at multiple input resolutions, with hybrid, class weighted loss functions to better capture small but clinically relevant tumor components. Experiments quantify trade-offs between segmentation quality and computational cost, reporting not only Dice based performance but also memory usage, training time and approximate cloud compute expenses. Results show strong performance on pre-treatment data but marked degradation in post-treatment cases, particularly for enhancing core and resection cavity, and link typical failures to specific imaging patterns such as fragmented cavities and treatment related changes. We recommend configurations that balance accuracy and cost for practical deployment and outline directions for improving robustness in post treatment segmentation.