Decoupled I/O-Dominant Pipelines for Large-Scale Whole-Slide Image Embedding Extraction

Mayanka Chandrashekar; Xi Zhang; Ethan Seefried; Tirthankar Ghosal; John Gounley; Heidi Hanson

cs.DCAug 27, 2026

Decoupled I/O-Dominant Pipelines for Large-Scale Whole-Slide Image Embedding Extraction

Authors: Mayanka Chandrashekar, Xi Zhang, Ethan Seefried, Tirthankar Ghosal, John Gounley, Heidi Hanson

Organizations: Oak Ridge National Laboratory

Abstract

Whole-slide images (WSIs) are central to computational pathology but are prohibitively large, making patch-based processing the practical unit for foundation model inference. At scale, however, generating and handling massive numbers of patches on quickly introduces significant I/O and orchestration overhead, often dominating end-to-end performance. We present a decoupled, I/O-aware pipeline for large-scale WSI embedding extraction that decomposes the workflow into three stages: (1) patch generation and staging, (2) embarrassingly parallel embedding inference, and (3) sharded vector database ingestion. This design isolates data movement from compute, enabling efficient patch delivery, scalable multi-node inference with minimal communication. The resulting system produces a distributed vector database where embeddings are persistently coupled with rich metadata (e.g., patient, slide, and patch attributes), enabling efficient filtering, retrieval, and downstream reuse. This representation database is compact and reusable for tasks such as retrieval, classification, and few-shot learning, particularly benefiting low-resource environments. We show that decoupling I/O, computation, and ingestion enables high-throughput WSI embedding extraction at scale. By characterizing the scaling envelope, we demonstrate that storage dominates beyond moderate concurrency, reframing WSI embedding extraction as a data-centric systems problem rather than a purely compute-bound workload.

Figures & tables

Category	Model / System	Learning Type	Data Scale	Key Strength	Pipeline Coupling	I/O Awareness	Scalability	Embedding Reuse
WSI Representation Models Used in This Work
Hierarchical	HIPT [ 2 ]	Hierarchical SSL	10,678 WSIs	Multi-scale tissue modeling	N/A	$\times$	N/A	Partial
Foundation Model	Virchow / Virchow2 [ 3 ]	Large-scale SSL	1.5M–3.1M+ H&E slides	Strong general-purpose features	N/A	$\times$	N/A	Partial
Foundation Model	H-Optimus / H-Optimus-0 [ 4 ]	Pathology FM	Large-scale histopathology corpus	Pathology-specialized representation learning	N/A	$\times$	N/A	Partial
WSI and HPC Pipelines
Traditional Pipelines	OpenSlide-based workflows [ 5 ]	Patch extraction + inference	Variable	Simple and widely used	Tight	$\times$	Limited	$\times$

TABLE I : Comparison of representative WSI models, WSI/HPC pipelines, and data systems. Modern pathology models improve representation quality but generally assume efficient data access. Our work targets the orthogonal systems problem: scalable, I/O-aware embedding extraction with persistent reuse.

Fig. 1 : Decoupled WSI embedding pipeline. The workflow is organized into three stages: (1) MPI-based patch generation and staging, (2) embarrassingly parallel embedding inference across GPUs, and (3) shard-parallel vector database ingestion. Each stage is parallelized according to its dominant constraint—storage, compute, and write throughput—enabling independent scaling. The pipeline produces a persistent, metadata-aware embedding database for downstream reuse.

CCDI Dataset
No. of Tissue Samples	4185
No. of Patients	4054
No. of DICOMS	19,603
No. of Patches (256x256)	414M
No. of Non-White Patches	170M

TABLE II : CCDI Dataset Statistics

Slide	Patch Size	Nodes	MPI Ranks	Runtime (s)	Files Created	Output Size (GB)	Avg File Size (KB)	Write Throughput (MB/s)
ff93fcc3	224	1	56	19.82	93,696	2.42	25.8	116.22
ff93fcc3	256	1	56	17.21	71,773	2.40	33.4	133.03
d47c2741	224	1	56	40.67	184,785	6.75	36.5	158.20
d47c2741	256	1	56	34.11	141,561	6.71	47.4	187.60
ca6e4e7a	224	1	56	47.24	224,133	9.94	44.4	200.67
ca6e4e7a	256	1	56	42.02	171,702	9.88	57.5	224.27

TABLE III: Stage 1 patch-generation performance for three representative slides at 56 MPI ranks using patch sizes 224 and 256. The selected slides span smaller, medium, and larger output footprints in the benchmark set. Increasing patch size reduces the number of generated files while increasing average file size and effective write throughput.

Patch Size	Nodes	MPI Ranks	Runtime (s)	Speedup	Efficiency (%)	Files Created	Write Throughput (MB/s)
224	1	56	22.39	1.00 $\times$	100.0	93,696	102.66
224	2	112	18.85	1.19 $\times$	59.4	93,696	122.99
224	4	224	13.22	1.69 $\times$	42.3	93,696	177.50
256	1	56	16.89	1.00 $\times$	100.0	71,773	135.46
256	2	112	13.41	1.26 $\times$	63.0	71,773	171.57
256	4	224	9.56	1.77 $\times$	44.2	71,773	244.02

TABLE IV: Stage 1 strong-scaling results for one representative slide (ID ff93fcc3-7bc2-4656-8e2d-e379180fa042 ) using patch sizes 224 and 256 across 56, 112, and 224 MPI ranks. Speedup and parallel efficiency are computed relative to the 56-rank baseline for each patch size. The sublinear scaling and rising write throughput are consistent with the I/O-dominated behavior

Panel of Figure 6 — Fig. 2 : Setup A (Patch Sweep). As patch count increases, throughput plateaus while I/O wait grows and GPU utilization remains modest, indicating a transition from compute-bound to I/O-dominated (storage-limited) execution. This shows that increasing workload size does not improve accelerator utilization when data delivery becomes the bottleneck.

Fig. 3 : Setup B. Strong-scaling efficiency versus GPU count across models. Despite fully independent execution with no collective communication in the critical path, scaling deviates from linear and efficiency drops at higher concurrency. This demonstrates that performance is limited by shared storage bandwidth rather than communication overhead.

Model	GPUs	Throughput	Speedup	Efficiency	I/O %
h-optimus-0	1	110.7	1.00x	100.0%	1.8
h-optimus-0	2	187.7	1.70x	84.8%	2.5
h-optimus-0	4	284.0	2.57x	64.1%	3.5
h-optimus-0	8	362.5	3.27x	40.9%	6.5
h-optimus-0	16	600.3	5.42x	33.9%	10.7
h-optimus-0	40	855.8	7.73x	19.3%	15.5

TABLE V : Strong scaling of embedding inference across models. Throughput increases with GPU count, but efficiency degrades and I/O fraction rises with concurrency. This indicates that scaling is limited by shared storage bandwidth rather than compute or communication.

Panel of Figure 9 — Fig. 4 : Setup C (Kernel-level analysis)

Panel of Figure 10 — Fig. 5 : Setup D (Production Scale Behavior)

Model	Total Rows	Avg Rate	Max Rate	Files	Mean File Time (s)	Total Time (min)	RSS Max (GB)
HIPT	6,143,506	2628.88	2775	14,499	4.27	1032.18	0.62
Optimus	2,298,630	258.71	321	4,052	55.59	3754.24	1.68
Virchow2	5,413,012	131.86	155	11,726	116.52	22772.80	2.74

TABLE VI: Performance comparison of large-scale VDB ingestion of embedding across three foundation models. Rows are sorted by average ingestion throughput (rows/sec), highlighting substantial performance differences across embedding backbones. HIPT achieves the highest throughput and lowest per-file processing time, while Virchow2 exhibits significantly slower ingestion despite processing a comparable number of rows. We also observe increased memory footprint and total runtime for larger embedding sizes (e.g., Virchow2), emphasizing the importance of model efficiency in distributed vector database construction.

Nodes	Agg. Rate (rows/s)	Wall Time (min)	Speedup	Efficiency (%)	RSS Max (GB)
1	22,216 ∗	127.9 ∗	1.00 $\times$	100.0	0.58
2	40,694	83.84	1.53 $\times$	76.3	0.59
4	86,099	36.90	3.47 $\times$	86.7	0.60
8	188,020	20.79	6.15 $\times$	76.9	0.59

TABLE VII: Strong scaling ablation for HIPT embedding VDB ingestion on Frontier. Each configuration ingests the same

{\sim}170

M embedding rows into Milvus; we vary the number of nodes (16 ranks each) and report aggregate throughput, wall-clock time, speedup, and parallel efficiency (speedup

/N

). ∗ Extrapolated from observed rate; the 1-node run did not complete within the 2-hour allocation. The 4-node configuration achieves the highest parallel efficiency (

86.7\%

) and corresponds to the production run in Table VI .

Whole-Slide Image Analysis

Whole-slide image analysis uses AI to interpret gigapixel pathology slides for diagnosis, prognosis, tissue mapping, and report generation.

See all 214 papers

Masked Feature Encoding for Large-Scale Whole Slide Image Representation
Oct 7, 2026Haoyu He, Basile Tessier-Cloutier, Yang Wang +1Whole-Slide Image Analysis Medical Image Representation Learning
Department of Computer Science and Software Engineering (CSSE), Concordia University, Montreal, Canada · McGill University, Montreal, Canada · Mila – Quebec AI Institute, Montreal, Canada
Domain Adaptation Without the Compute Burden for Efficient Whole Slide Image Analysis
Mar 16, 2026Umar Marikkar, Muhammad Awais, Sara AtitoWhole-Slide Image Analysis Computational Pathology
Institute for People-Centered AI, University of Surrey · Centre of Vision, Speech and Signal Processing, University of Surrey
Thinking in Scales: Accelerating Gigapixel Pathology Image Analysis via Adaptive Continuous Reasoning
May 19, 2026Jiusong Ge, Yingkang Zhan, Wenjie Zhao +13Computational Pathology Adaptive Inference
School of Computer Science and Technology, Xi’an Jiao-tong University, Xi’an, China · Department of Transmedia Art, Xi’an Academy of Fine Arts, Xi’an, China · Language Technology Lab, University of Cambridge, Cambridge, U.K. +3

Oct 7, 2026J/K move · Enter open · S save

Masked Feature Encoding for Large-Scale Whole Slide Image Representation

Haoyu He, Basile Tessier-Cloutier, Yang Wang, Mahdi S. Hosseini

Department of Computer Science and Software Engineering (CSSE), Concordia University, Montreal, Canada · McGill University, Montreal, Canada · Mila – Quebec AI Institute, Montreal, Canada

Whole-Slide Image Analysis Medical Image Representation Learning Histopathology Image Classification Multiple Instance Learning Masked Image Modeling

Open paper AskPDF arXiv

Whole slide image (WSI) analysis in computational pathology follows a multiple instance learning (MIL) pipeline where patch embeddings are extracted independently and aggregated for slide-level prediction, but within-slide variance from staining, scanner, and local texture can overwhelm the discriminative signal. We propose Masked Feature Encoding for Multiple Instance Learning (MFE-MIL), a feature-space masking framework that trains a lightweight MLP adapter jointly with a window-based masked reconstruction branch and a MIL classification head. The two objectives are complementary. Classification guides the adapter to suppress within-slide patch variance, while window-based masked reconstruction provides an auxiliary regularizer for the adapted features without using patch coordinates, coordinate graphs, or segmentation preprocessing. The raster patch-extraction order is used only as a weak implicit prior. At inference, the decoder is removed, leaving only the adapter and MIL head. Across CAMELYON16/17, PANDA, and TCGA-BRCA with four diverse encoders, MFE-MIL improves ACC/F1 for nearly all tested aggregator-encoder settings and AUC in most, outperforms coordinate-based spatial methods (CAMIL), and achieves higher AUC than 2DMamba on three of four datasets (UNI). On five TCGA survival cohorts it improves the average concordance index for every aggregator tested, its most consistent gain. Code is available at https://github.com/AtlasAnalyticsLab/MFE-MIL.

Abstract

Whole-Slide Image Analysis

Explore similar work

Masked Feature Encoding for Large-Scale Whole Slide Image Representation

Domain Adaptation Without the Compute Burden for Efficient Whole Slide Image Analysis

Thinking in Scales: Accelerating Gigapixel Pathology Image Analysis via Adaptive Continuous Reasoning