MoA: Mixture of Aggregators Improves Slide-Level Diagnosis in Computational Pathology

Fatih Ozlugedik, Muhammed Furkan Dasdelen, Rao Muhammad Umer, Carsten Marr
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:2757-2779, 2026.

Abstract

Multiple instance learning (MIL) is the standard for learning slide-level representations from whole slide images (WSIs), typically using a single attention-based aggregator to pool instance features. However, a single aggregator can struggle to capture morphological and compositional patterns of cells in pathology and cytology data, and different diseases may demand different pooling behaviours. We propose a mixture-of-aggregators framework that models complementary aspects of instance distributions in histology and hematologic cytology. A router with top-2 gating dynamically selects the most relevant aggregators per slide, and their outputs are fused into a patient-level representation. To avoid collapse to a single dominant expert aggregator, we add a load-balancing loss and Gumbel noise on the router logits. We evaluate our method on 19 different tasks from 16 datasets including histology and hematologic cytology. Compared to single-aggregator baselines, our approach improves diagnostic prediction accuracy by an average of 4.5% over ABMIL and 12.6% over TransMIL across all tasks. Beyond performance, our analysis shows that different aggregators attend to distinct, disease-specific instance distributions, providing interpretable insights into the diagnostic process.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-ozlugedik26a, title = {MoA: Mixture of Aggregators Improves Slide-Level Diagnosis in Computational Pathology}, author = {Ozlugedik, Fatih and Dasdelen, Muhammed Furkan and Umer, Rao Muhammad and Marr, Carsten}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {2757--2779}, year = {2026}, editor = {Huo, Yuankai and Gao, Mingchen and Kuo, Chang-Fu and Jin, Yueming and Deng, Ruining}, volume = {315}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v315/main/assets/ozlugedik26a/ozlugedik26a.pdf}, url = {https://proceedings.mlr.press/v315/ozlugedik26a.html}, abstract = {Multiple instance learning (MIL) is the standard for learning slide-level representations from whole slide images (WSIs), typically using a single attention-based aggregator to pool instance features. However, a single aggregator can struggle to capture morphological and compositional patterns of cells in pathology and cytology data, and different diseases may demand different pooling behaviours. We propose a mixture-of-aggregators framework that models complementary aspects of instance distributions in histology and hematologic cytology. A router with top-2 gating dynamically selects the most relevant aggregators per slide, and their outputs are fused into a patient-level representation. To avoid collapse to a single dominant expert aggregator, we add a load-balancing loss and Gumbel noise on the router logits. We evaluate our method on 19 different tasks from 16 datasets including histology and hematologic cytology. Compared to single-aggregator baselines, our approach improves diagnostic prediction accuracy by an average of 4.5% over ABMIL and 12.6% over TransMIL across all tasks. Beyond performance, our analysis shows that different aggregators attend to distinct, disease-specific instance distributions, providing interpretable insights into the diagnostic process.} }
Endnote
%0 Conference Paper %T MoA: Mixture of Aggregators Improves Slide-Level Diagnosis in Computational Pathology %A Fatih Ozlugedik %A Muhammed Furkan Dasdelen %A Rao Muhammad Umer %A Carsten Marr %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-ozlugedik26a %I PMLR %P 2757--2779 %U https://proceedings.mlr.press/v315/ozlugedik26a.html %V 315 %X Multiple instance learning (MIL) is the standard for learning slide-level representations from whole slide images (WSIs), typically using a single attention-based aggregator to pool instance features. However, a single aggregator can struggle to capture morphological and compositional patterns of cells in pathology and cytology data, and different diseases may demand different pooling behaviours. We propose a mixture-of-aggregators framework that models complementary aspects of instance distributions in histology and hematologic cytology. A router with top-2 gating dynamically selects the most relevant aggregators per slide, and their outputs are fused into a patient-level representation. To avoid collapse to a single dominant expert aggregator, we add a load-balancing loss and Gumbel noise on the router logits. We evaluate our method on 19 different tasks from 16 datasets including histology and hematologic cytology. Compared to single-aggregator baselines, our approach improves diagnostic prediction accuracy by an average of 4.5% over ABMIL and 12.6% over TransMIL across all tasks. Beyond performance, our analysis shows that different aggregators attend to distinct, disease-specific instance distributions, providing interpretable insights into the diagnostic process.
APA
Ozlugedik, F., Dasdelen, M.F., Umer, R.M. & Marr, C.. (2026). MoA: Mixture of Aggregators Improves Slide-Level Diagnosis in Computational Pathology. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:2757-2779 Available from https://proceedings.mlr.press/v315/ozlugedik26a.html.

Related Material