arXiv subjects

Piotr Krajewski

Publications and source records attributed to Piotr Krajewski.

3 recordsLinked to original sources

arXiv2026

Distribution-based deep multiple instance learning for tumor proportion scoring in NSCLC

Accurate assessment of tumor proportion score (TPS) in non-small cell lung cancer (NSCLC) is critical for treatment planning and prognosis. Key challenges include the tedious manual work required to annotate each slide, combined with the limited number of experts certified for this task. Multiple instance learning (MIL) has proven to be an effective approach for predicting TPS scores at the slide level; however, existing methods struggle with non-expressive (zero class) images. Our approach involves two models: (1) an embedding-extraction and multiclass-classification network that captures the histopathological features of individual patches, and (2) a MIL model that aggregates these embeddings to predict zero-inflated beta (ZIBeta) parameters representing the overall TPS probability distribution for the entire slide. Using only slide-level TPS scores as labels, we demonstrate how this end-to-end framework can leverage a novel distribution-based architecture to improve prediction accuracy and explainability. ZIBeta modeling significantly outperforms baseline linear and ridge regression while capturing expected accuracy through distribution concentration.

cs.CV↗

arXiv2025

RF-DETR for Robust Mitotic Figure Detection: A MIDOG 2025 Track 1 Approach

Mitotic figure detection in histopathology images remains challenging due to significant domain shifts across different scanners, staining protocols, and tissue types. This paper presents our approach for the MIDOG 2025 challenge Track 1, focusing on robust mitotic figure detection across diverse histological contexts. While we initially planned a two-stage approach combining high-recall detection with subsequent classification refinement, time constraints led us to focus on optimizing a single-stage detection pipeline. We employed RF-DETR (Roboflow Detection Transformer) with hard negative mining, trained on MIDOG++ dataset. On the preliminary test set, our method achieved an F1 score of 0.789 with a recall of 0.839 and precision of 0.746, demonstrating effective generalization across unseen domains. The proposed solution offers insights into the importance of training data balance and hard negative mining for addressing domain shift challenges in mitotic figure detection.

eess.IV↗

arXiv2025

Foundation Model-Driven Classification of Atypical Mitotic Figures with Domain-Aware Training Strategies

We present a solution for the MIDOG 2025 Challenge Track~2, addressing binary classification of normal mitotic figures (NMFs) versus atypical mitotic figures (AMFs). The approach leverages pathology-specific foundation model H-optimus-0, selected based on recent cross-domain generalization benchmarks and our empirical testing, with Low-Rank Adaptation (LoRA) fine-tuning and MixUp augmentation. Implementation includes soft labels based on multi-expert consensus, hard negative mining, and adaptive focal loss, metric learning and domain adaptation. The method demonstrates both the promise and challenges of applying foundation models to this complex classification task, achieving reasonable performance in the preliminary evaluation phase.

eess.IV↗