aiqbench analysis
The paper introduces a flow-matching transformer that generates whole-slide pathology images at up to 32,768×32,768 pixels, using a sliding-window inference pipeline and an adaptive scheduler that gives a reported 3x speedup. The most transferable result is the reported gain in few-shot slide classification (up to 18%) from augmenting training data with synthetic samples, though that comes from the authors' benchmarks and needs independent validation. Readers should also note that generative quality metrics for medical imagery don't by themselves establish clinical utility.
BigCell: Generating Gigapixel Whole-Slide Images
arXiv:2610.14261v1 Announce Type: cross Abstract: Whole-slide images (WSIs) are gigapixel-scale digital scans of histopathology tissue and serve as the primary basis for cancer diagnosis. Existing histopathology generative models operate at the patch scale, producing fixed-size tiles of at most 1024$\times$1024 pixels which encompass only a small fraction of the WSI. In contrast, text reports for clinical diagnoses and slide-level diagnostic labels -- the supervision that exists at scale -- operate at the whole-slide level. We present BigCell, the first approach to generate coherent whole-slide histopathology images at native gigapixel resolution. BigCell trains a flow-matching transformer at 2048$\times$2048 pixels on over 120,000 publicly available WSIs spanning multiple organ types. Conditioning on embeddings from pathology foundation models and using an inference-time sliding-window pipeline, BigCell generates coherent images of up to 32,768$\times$32,768 pixels, outperforming prior tile generators on both tile-level and slide-level fidelity metrics. To make gigapixel synthesis practical, we introduce an adaptive scheduling strategy that prioritizes important slide regions, speeding up image generation by over 3\ttimes with minimal quality degradation. Furthermore, we propose the first text-to-image synthesis at 8k resolution by training a separate model that produces dense conditioning grids directly from pathology text reports. We demonstrate that augmenting with synthetic samples improves slide-level few-shot classification performance by up to $18 \%$ over baselines trained on real data.
Authors:Srikar Yellapragada, Alexandros Graikos, Zilinghan Li, Kostas Triaridis, Tarak Nath Nandi, Ji Dong K Bai, Beatrice Knudsen, Tahsin Kurc, Prateek Prasanna, Rajarsi R. Gupta, Ravi K Madduri, Joel Saltz, Dimitris Samaras
Abstract:Whole-slide images (WSIs) are gigapixel-scale digital scans of histopathology tissue and serve as the primary basis for cancer diagnosis. Existing histopathology generative models operate at the patch scale, producing fixed-size tiles of at most 1024$\times$1024 pixels which encompass only a small fraction of the WSI. In contrast, text reports for clinical diagnoses and slide-level diagnostic labels -- the supervision that exists at scale -- operate at the whole-slide level. We present BigCell, the first approach to generate coherent whole-slide histopathology images at native gigapixel resolution. BigCell trains a flow-matching transformer at 2048$\times$2048 pixels on over 120,000 publicly available WSIs spanning multiple organ types. Conditioning on embeddings from pathology foundation models and using an inference-time sliding-window pipeline, BigCell generates coherent images of up to 32,768$\times$32,768 pixels, outperforming prior tile generators on both tile-level and slide-level fidelity metrics. To make gigapixel synthesis practical, we introduce an adaptive scheduling strategy that prioritizes important slide regions, speeding up image generation by over 3\ttimes with minimal quality degradation. Furthermore, we propose the first text-to-image synthesis at 8k resolution by training a separate model that produces dense conditioning grids directly from pathology text reports. We demonstrate that augmenting with synthetic samples improves slide-level few-shot classification performance by up to $18 \%$ over baselines trained on real data.
Accepted at NeurIPS 2026. Project page: this https URL
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2610.14261 [cs.CV]
(or arXiv:2610.14261v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.14261 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 275 words
Related coverage
Biomedical Imaging's Real Bottleneck Is the Data, Not the Model
Databricks · Oct 8, 2026
GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models
Microsoft Research · Aug 31, 2026
A Multimodal Agentic Pathology Co-pilot via Evidence Grounded Reasoning
arXiv · Aug 19, 2026
A Multi-modal Agentic Co-pilot for Evidence Grounded Computational Pathology
arXiv · Jun 9, 2026