NY · NJ · PA · OH · New England — research basis, September 2026

Northeast Imaging Map

Which hospitals and academic centers in the Northeast and Ohio already move digital images across scanners, PACS and institutions, what estates they run, and which funding routes reach them without a hospital IT capital request. Prepared for Tiger Health Software as a vendor-neutral orchestration layer entering the US as non-diagnostic infrastructure.

Status limited to public documentation [S] sourced · [D] derived · [I] inference
Framing

What the research supports

The formats are already fragmented

The multi-site pathology programs in this region that publish their infrastructure do not agree on a format. NCI's Imaging Data Commons converts whole slide images to DICOM Slide Microscopy on Google Cloud and AWS. The Cancer Imaging Archive's histopathology portal stores native SVS, NDPI and TIFF. HTAN's primary format is OME-TIFF with SVS accepted, on Synapse over S3 and GCS. The cooperative-group biobanks in Ohio and Pennsylvania that scan at volume disclose no format at all.

A layer whose function is to move images between those systems without committing the institution to one of them has a defined place in that landscape. That is a description of the problem, not evidence that any institution has budgeted to solve it.

Non-diagnostic entry sidesteps the two slowest gates

Primary-diagnosis pathology software in the US sits behind FDA clearance and, at the hospital, behind the value analysis committee, which is triggered by PHI and LIS integration rather than by dollar value and takes three to twelve months. Research, archive, tumor-board and cross-site collaboration use cases avoid both.

The cost of that entry point is that it lands in research budgets rather than clinical ones, which are smaller, grant-cycled, and controlled by cancer-center and department leadership rather than by hospital operations. The funding instruments below are sized accordingly. This research treats a $30K ARR as a vice-president-level decision by assumption, not by verification: no institution's director-level signature threshold was confirmed, and no public benchmark for US hospitals exists. The gate map's own inferred director ranges reach $100K at the largest systems, so a director-level close at $30K is possible at some of them and unverified at all of them.

Digital pathology is additive cost. The lab still cuts and stains glass. The 43 Category III CPT codes carry no national RVUs and are packaged under OPPS. US lab adoption is roughly 10% (CAP, 2024). UPMC's own published all-in figure is $12–15 per slide, split evenly across hardware, software, technician labor and storage. That is why the capital committee declines, and why every route on this page runs around it.
Institutions

Institutions, by evidence of adoption

Pathology is what is publicly documented for whole slide imaging. Imaging estate is the radiology PACS and archive stack where it could be found; "undisclosed" means the viewer contract is public but the archive vendor is not. Greenfield means a search found no deployment, pilot or program. Revenue is the most recent published fiscal year; NIH is FY2025; CCSG is the FY2025 P30 obligation from HHS TAGGS. Systems under roughly $1.5B in revenue appear only when they are among the largest in their state.

State
Pathology status
Signal
Search
Institution St Pathology Imaging estate Signals Revenue NIH FY25 CCSG FY25 Funding route
Deployed production clinical use Partial live but narrow Intent named priority or active procurement Research core or lab only Greenfield no evidence found
Multi-vendor imaging estate confirmed or likely multi-vendor Multi-site research lead or member in a cross-institution image program AI investment named AI center, compute or partnership with figures Timing procurement, migration or grant cycle live in the next 12 months Fujifilm Fujifilm product or project documented; mentions highlighted in the row

Fujifilm on this page

  • Hackensack Meridian Health (NJ). Synapse Radiology PACS (Synapse 5) with Synapse VNA and Synchronicity, Synapse Mobility viewer and Synapse Cloud Services purchased 2017; network-wide Synapse Cardiology PACS across 18 hospitals announced March 4, 2025, described as cloud-ready. The only confirmed end-to-end single-vendor Fujifilm estate in the region.
  • OhioHealth (OH). Synapse PACS since 2003 and Synapse Mobility from 2011; no newer public record, so whether the estate is current or overdue could not be established.
  • Capital Health (NJ), not in the table. Synapse PACS replacement, August 2017, one of seven systems in a single Fujifilm announcement.
  • Market position (KLAS). Synapse VNA was Best in KLAS for VNA in 2024 at 86.7 against Merative iConnect at 80.3. In KLAS's 2025 large-PACS scores Fujifilm sits third at 84.9, behind Sectra 91.0 and Agfa 87.2. KLAS's Nov 2023 report notes VNA adoption is strongest when bundled with the PACS vendor, naming Fujifilm as the example.
  • Not Fujifilm. "Synapse" in the HTAN row and the Dana-Farber note is Sage Bionetworks' Synapse research data platform, unrelated to Fujifilm Synapse. Those mentions are deliberately not highlighted.
  • Not found. No Fujifilm digital pathology deployment at any institution in the region, and no Fujifilm reference at any of the 34 institutions whose radiology stack could not be established.
Research that moves images

Programs in the footprint that depend on shared imaging

Funded medical, public-health and population research with a lead or member institution in the ten states, grouped by domain. The infrastructure column is the point: it says whether images actually leave the institution, on what platform, and in what format. A tag at the start of each infrastructure cell classifies the movement. Centralized means images are copied to a shared repository; Federated means models move and images stay; Single-site means one institution's own holdings; Data-only means the program shares structured data and no images; Undisclosed means the program publishes nothing about its image platform.

What the pattern says

  • The centralized programs each built their own pipeline. IDC converts to DICOM on Google Cloud and AWS; ISIC sits in an AWS S3 bucket; KPMP's atlas is on AWS; ADNI, SCAN and ABCD route through LONI or the NIH data hubs; CBTN puts MRI in Flywheel; TMIST used ACR TRIAD; NIOSH mandates DICOM transfer by regulation. None of these is a candidate to replace its pipeline. They are evidence that the movement problem is real and gets solved one program at a time.
  • The gaps are stated in the programs' own words. UPMC's diabetic retinopathy program stores and forwards images through Epic inside UPMC and by secure email from non-UPMC sites. The World Trade Center Health Program has eight clinical centers of excellence in the footprint and no cross-center imaging archive. NACC's digital neuropathology stream is a pilot with a test dataset. The cooperative-group biobanks at Nationwide, Pittsburgh and Ohio State scan at volume and disclose no platform. Rutgers' registry-linked pathology framework centralizes whole slide images across state lines using Box for bulk transfer.
  • The institutions that recur are a short list. MSK (HTAN DCC, CAIA, ISIC, QIN, Warren Alpert Center), Penn and CHOP (CBTN, Kids First, HuBMAP, CureGN, FeTS lineage, QIN, e-ROP, PMBB), Pittsburgh (KPMP, ADNI, ABCD, BDSA, SenNet CODCC, NRG, FeTS), Yale (KPMP recruitment and tissue, ADNI, ABCD, HuBMAP, SenNet, HTAN Phase 2, SEER VTR via the Connecticut Tumor Registry), Columbia (KPMP, ADNI, CureGN, HuBMAP, SenNet), MGH and Brigham (IDC, HCP, Alliance, HTAN, FL4M, CHoRUS, NHS tissue), Mount Sinai (KPMP hub, ADNI, WTC data center, I-ELCAP, AIR·MS), Stony Brook (TCIA, QuIP registry platform, FeTS).
  • Public health is data-only more often than not. The two Breast Cancer Surveillance Consortium registries in the footprint (Vermont, New Hampshire) and the ACR national registries hold structured data, not images. The exceptions that do move images are occupational (NIOSH) and screening programs run by a single system (UPMC, NYC H+H).

Oncology tissue and slides

ProgramRegional institutionsFundingImage infrastructure
NCI Imaging Data CommonsBrigham and Women's (co-PIs Fedorov, Kikinis); MGHNCI via Cancer Moonshot, contract through Leidos Biomed; amount undisclosedCentralized
DICOM Slide Microscopy for all WSI; Google Cloud and AWS public buckets; open-source SliM viewer
Cancer Imaging Archive, histopathology portalStony Brook (Saltz group, TIL maps for 13 cancer types); Frederick National LabNCI Cancer Imaging Program; since 2010Centralized
Native SVS, NDPI, TIFF; IDC is the DICOM front end; ACRIN and QIN legacy collections in DICOM
Human Tumor Atlas NetworkDCC at Dana-Farber and MSK. Phase 1 centers: HTAPP (Broad/DFCI/HMS/MGH), Pre-Cancer Atlas Pilot (BU), CHOP, Boston HTAN Center (DFCI/Broad/HMS/MIT), PATCH (HMS), Gerry Metastasis Center (MSK). Phase 2: DFCI myeloma, Yale lymphomaNCI U24CA233243 (DCC); amounts unpublished; Phase 2 from fall 2024Centralized
OME-TIFF primary, SVS accepted; Synapse (Sage) over S3 and GCS; DICOM subset to IDC; 10,608 biospecimens across 15 atlases
Cancer AI AllianceMSK and Dana-Farber, with Fred Hutch and Johns Hopkins$40M pooled; AWS, Google, Microsoft, NVIDIA partners; live Oct 2025Federated
Centralized orchestration layer on AWS; framework unnamed; pathology WSI not publicly listed as a modality
Registry-linked digital pathology frameworkRutgers Cancer Institute (Foran) with the NJ State Cancer Registry; NY State Cancer Registry; platform hosted at Stony BrookNIH UG3/UH3 CA225021, U24 CA215109, U24 CA180924, UL1 TR003017; NSF XSEDE; amounts unstatedCentralized
Olympus VS120; .vsi and .svs; WSIs and features on QuIP at Stony Brook; Box for bulk transfer; Oracle warehouse; Google Cloud tested. 772 prostate, 1,410 NSCLC, 268 breast, 48 lymphoma cases across state lines (2022)
SEER Virtual Tissue Repository pilotConnecticut Tumor Registry (CT DPH, Yale collaborates); NJ and NY registries were not in the pilotNCI SEER contracts plus a PanCAN donation; amounts unstatedCentralized
H&E slides scanned at a central laboratory; 175 pancreatic and 196 breast patients processed; scanner, format and hosting undisclosed; a searchable web interface is a stated goal, not built
NRG Oncology Biospecimen BankColumbus (Nationwide Children's) and Pittsburgh sitesNCI U24CA196067Undisclosed
51,333 slides scanned across 32 trials through Dec 2022; scanner and platform not disclosed
Alliance Biorepository SystemOhio State and Brigham and Women's, with WashU and MayoNCI NCTNUndisclosed
Image-analysis committee exists; infrastructure not disclosed
Children's Oncology Group Biopathology CenterNationwide Children'sNCIUndisclosed
Digital pathology review with slide-scanning robots; vendor and volumes not disclosed
Nurses' Health Study and HPFS tumor tissueHarvard Chan and BWH Channing; DF/HCC Tissue Microarray and Imaging CoreNCI cohort grants; tissue-specific awards not foundCentralized
Tumor blocks retrieved from hospitals nationwide to Boston; Leica scanners at the DF/HCC core; one nested study digitized 3,795 WSIs; hosting undisclosed
MSK Warren Alpert CenterMSKWarren Alpert Foundation, est. 2017Single-site
In-house HoBBiT de-identification and transfer tool; dedicated compute; ~7M slides licensed to Paige, now Tempus, on Azure
NCI Quantitative Imaging NetworkCHOP (with Children's National), Thomas Jefferson, Weill Cornell, NYU with MSKNCI U01 per team; program winding down, final funding 2027Centralized
Historically shared through TCIA in DICOM; current infrastructure not on the program page

Neurology and neuroimaging

ProgramRegional institutionsFundingImage infrastructure
ADNI4BIDMC, Boston University, Brigham and Women's; Albany Medical College, Columbia, Mount Sinai, Nathan Kline, NYU, Rochester; Penn, Pittsburgh; Yale; Butler Hospital, Rhode Island Hospital; Case Western/UH Cleveland, Ohio StateNIA public-private partnership; amount unstated; enrolling since 2023Centralized
MRI, amyloid and tau PET to the LONI Image Data Archive at USC; up to 1,500 subjects
ABCD StudyPittsburgh, Rochester, Vermont, Yale (of 21 sites)NIH, NIDA-led; amount unstatedCentralized
~11,878 participants; DICOM and BIDS; access moved June 2025 from the NIMH Data Archive to the NIH Brain Development Cohorts Data Hub, compute-in-place
Human Connectome Project LifespanMGH (Aging), Harvard (Development), with WashU, Minnesota, UCLA, OxfordNIH Blueprint; amount unstatedCentralized
NIfTI; HCP-A and HCP-D on the NIMH Data Archive; AABC Release 2 is 1,396 adults and 42+ TB on ConnectomeDB (Jan 2026)
SCAN (Standardized Centralized ADRD Neuroimaging)All ADRCs submitting prospectively; in-footprint ADRCs include Penn, Pitt, Columbia, Mount Sinai, NYU, BU, MGH/Harvard, Yale, Case WesternNIA U24 AG067418; launched Jan 2021Centralized
MRI and PET uploaded to LONI IDA, de-identified and defaced there; derived measures via NACC
NACC digital neuropathology pilotPilot partners UC Davis and UW; goal is all 36 ADRCs (footprint list as above)NIA supplement to NACC; amount unstatedUndisclosed
Test dataset only; ingest planned through the NACC Data Platform on Flywheel; WSI metadata standard in development. Images not yet centralized.
Brain Digital Slide ArchiveUniversity of Pittsburgh (Pearce) co-leads with Emory, UC Davis, NorthwesternNIA/NINDS U24NS133949; amount not retrievedFederated
Open-source Digital Slide Archive (Girder/HistomicsTK); "databases hosted on multiple servers at different research institutions"; cloud vs on-prem unstated
FeTS / BraTSPenn CBICA lineage; coordination now at Indiana. Regional members: MGH, Brigham, NYU, Stony Brook, Mount Sinai, Penn, Pittsburgh, Jefferson, Case Western, Weill CornellNCI ITCR U01CA242871, U24CA189523; amounts unstated. fets.ai domain has lapsedFederated
Intel OpenFL orchestration, GaNDLF training; data stays on site; NIfTI registered to SRI-24, skull-stripped, pseudo-identified; 2024 challenge drew 1,251 training cases from 23 institutions
Framingham Heart Study Brain Aging ProgramBoston UniversityNIA U19 AG068753, $26.56M over five years (2020)Single-site
Brain MRI in native DICOM since 1999, cardiac MRI 2002–2006; access via FHS-BAP data portal; storage platform undisclosed
FL4M federated consortiumMGH lead; LaheyNot disclosedFederated
Rhino Health platform; radiology (aneurysm)

Nephrology, dermatology, pediatrics

ProgramRegional institutionsFundingImage infrastructure
Kidney Precision Medicine ProjectCentral hub includes Mount Sinai. Recruitment: Boston University, Cleveland Clinic, Columbia, Joslin/BIDMC, Mount Sinai, Pittsburgh, Yale. Tissue interrogation: Ohio State (with Indiana), YaleNIDDK; ~$3.7M/yr recruitment sites, ~$5M/yr tissue sites, ~$4M/yr hub, ~$2M/yr atlas center, five-year periodsCentralized
Biopsy WSIs (H&E, PAS, trichrome, silver), EM; Atlas Repository at atlas.kpmp.org on AWS; format and de-identification undisclosed
NEPTUNE31 enrolling centers; regional list not foundNCATS/NIDDK U54DK083912; RDCRN; NephCureCentralized
Digital Pathology Repository at Michigan; WSIs scanned at 40x on Hamamatsu and Aperio; surfaced via tranSMART
CureGNParticipating clinical centers at Columbia and Penn (of four); 65 sites totalNIDDK UM1DK100845 and related; amounts undisclosedUndisclosed
Biopsy digital pathology implied by the shared DCC; platform not on the program site
ISIC ArchiveMSK is the lead institution (Rotemberg); UPMC Hillman contributorShore Family Fund; size undisclosedCentralized
549,571 public and 1,203,225 total dermoscopic, clinical and 3D images in an AWS S3 bucket (us-east-1); per-image CC licenses; 10,413 registered users
Children's Brain Tumor NetworkCHOP (D3b) lead. Members: UPMC Children's, NYU Hassenfeld, Weill Cornell, Westchester Medical Center, Hackensack Sanzari, Akron Children's, Dayton Children's (of 34)Foundation-supported; amount undisclosed; since 2011Centralized
7,986 enrolled; 29,119 clinical MRI sessions across 1,976 subjects in Flywheel; genomics on CAVATICA; histology slide count and WSI format undisclosed
Kids First Data Resource CenterCHOP lead, with Velsera, UChicago, Vanderbilt, UNC, Sainte-JustineNIH Common Fund U2CHL138346; amount undisclosedCentralized
53,000+ patients, 1.56M+ files (Feb 2026); imaging listed as a data type; CAVATICA (Velsera); cloud provider and imaging format not named
e-ROP retinopathy of prematurity studyPenn Center for Preventive Ophthalmology and Biostatistics (data center and reading center)NEI; completed ~2014Centralized
RetCam images from 12 NICUs uploaded to a study server at Penn; ~2,000 infants

Radiology datasets, trials and federated networks

ProgramRegional institutionsFundingImage infrastructure
fastMRINYU Langone CAI2R with MetaNIBIB P41EB017183; launched 2018, updated May 2025Single-site
Raw k-space in ISMRMRD, DICOM de-identified with the RSNA Clinical Trial Processor; ~1,500 knee and 6,970 brain raw cases plus 10,000 clinical DICOM; NYU under DUA and mirrored on AWS Open Data
TMIST (ECOG-ACRIN EA1151)145 sites, 411 mammography units; regional list not publishedNCI; closed to accrualCentralized
136,742 screening studies through 2022; de-identified and transmitted via ACR TRIAD in DICOM; archive location unstated
ACR Connect / AI-LAB13 US hospitals as of 2022; names not disclosedACRFederated
On-premise Linux appliance; DICOM from PACS and FHIR from EHR; anonymization profiles; model weights exchanged, images stay
MGB and Microsoft imaging foundation modelsMass General BrighamCommercial collaboration, July 2024; no figureUndisclosed
Azure AI platform; Nuance Precision Imaging Network as distribution; centralized vs federated unstated
Aidoc Diagnostic AI ConsortiumNorthwell, Mount Sinai, University Hospitals Cleveland, WellSpan among 12 (Aug 2026)Vendor-ledFederated
Aidoc aiOS; radiology; not an image archive
I-ELCAP lung screening consortiumMount Sinai coordinating center (Henschke, Yankelevitz)Simons Foundation International $12M over five years (2023) for the AIRS open-source triage systemUndisclosed
Web-based management system for member sites; whether CT DICOM is centralized is unstated

Tissue atlases

ProgramRegional institutionsFundingImage infrastructure
HuBMAPIntegration hub at CMU/Pittsburgh Supercomputing Center, Harvard Medical School, NY Genome Center. Tissue mapping centers: CHOP (two), Penn, Rochester, GE Global Research, BIDMC, UConn. Technology sites: Yale, Columbia, Broad, Harvard. Demonstration: Brigham, CHOPNIH Common Fund U54, UG3/UH3, OT2, U01; amounts unpublishedCentralized
Spatial and multiplexed tissue imaging on portal.hubmapconsortium.org; format and hosting not confirmed from pages that rendered
SenNetCoordination center at Pittsburgh. Human tissue centers: Yale, Pittsburgh, UConn, Columbia. Murine: Yale, Jackson Laboratory (Maine). Technology: Columbia, MITNIH Common Fund; UG3 to UH3 transition Sept 2024; amounts unpublishedCentralized
Portal at data.sennetconsortium.org (Pittsburgh); format and cloud undisclosed

Public health, surveillance and screening

ProgramRegional institutionsFundingImage infrastructure
Breast Cancer Surveillance Consortium registriesVermont Breast Cancer Surveillance System (UVM, Sprague); New Hampshire Mammography Network (Dartmouth, Tosteson); two of seven active registriesNCI; cycle 2022–2027; UVM cites P01 CA154292 and AI grants R01 CA262023, R37 CA292399Data-only
Structured exam, BI-RADS, pathology and registry linkage centralized at the coordinating center; the only image set is a 2005–2008 teaching dataset of 40,000 mammograms. UVM's AI grants use UVMMC images plus other registries' images through an undisclosed platform
ACR National Mammography Database and Lung Cancer Screening RegistryNational; facility lists by state not retrievedACR facility-fee registriesData-only
MQSA/BI-RADS and Lung-RADS data elements by manual entry, flat file or vendor transmission; no image submission
NIOSH Coal Workers' Health Surveillance ProgramNIOSH-approved facilities in PA and OH (map not enumerable); 184 B Readers nationally (Oct 2024)CDC/NIOSH appropriationCentralized
By regulation (42 CFR 37): facilities transmit DICOM chest radiographs to NIOSH within 14 days by secure transfer or portable media, then delete local copies; B Readers classify on calibrated workstations; anonymized images shared with Michigan State for AI research under a DUA
World Trade Center Health ProgramClinical centers of excellence: Mount Sinai (25,388 members), NYC H+H (19,716), NYU (18,375), Northwell (15,056), Stony Brook (14,899), FDNY, William Street, Rutgers (3,216); General Responder Data Center at Mount Sinai; 143,340 enrolled (Mar 2026)Federal (Zadroga Act); $348M medical claims in the past yearUndisclosed
Annual low-dose CT for eligible members; a Mount Sinai R21 analyzed 2,284 members' chest CTs. The data center manages health and exposure data; no cross-center imaging archive is described
UPMC point-of-care diabetic retinopathy program94 primary-care sites across UPMC and non-UPMC facilities in PennsylvaniaBeckwith Institute, departmental funds, local foundation camera grants; amounts unstatedCentralized
21,960 exams 2008–2020 on Centervue and Topcon cameras; UPMC sites store and forward through Epic, non-UPMC sites by secure email
NYC Health + Hospitals teleretinal screeningConey Island, Elmhurst, Lincoln, Queens, Woodhull; planned across all 11 acute hospitals and Gotham HealthNYS DOH Digital Healthcare Network grant; amount unstatedSingle-site
Non-mydriatic retinal photos read through the EMR within five business days; camera and archive undisclosed; 1,886 patients in the first four months

Biobanks and imaging warehouses

ProgramRegional institutionsFundingImage infrastructure
Penn Medicine BioBank Vision InitiativePenn Medicine (Witschey)UnstatedSingle-site
More than 875K CT, 500K MRI and 100K PET linked to 265K+ consented participants; Penn AInSights open-source AI orchestrator has screened 24K+ patients since May 2023; platform undisclosed
Mount Sinai AIR·MS and Imaging Research WarehouseIcahn School of Medicine at Mount SinaiNIH pilot for the warehouse; amount unstatedSingle-site
Radiology DICOM metadata for 2.06M patients, echo for 886K, pathology for 3.5M; on-prem Minerva HPC; de-identified image and EHR warehouse of 1M+ patients by REDCap request
Geisinger MyCodeGeisingerBiopharma partnerships; 370,000+ participantsUndisclosed
Genomic and EHR; imaging holdings not disclosed

Not found: TCGA/GDC slide-image infrastructure; any Labcorp or Quest digital consortium; Owkin or NVIDIA FLARE pathology pilots with regional sites; the 2025 DICOM WSI Connectathon participant list; TB chest X-ray programs at NYC, NJ or MA health departments; diabetic retinopathy tele-screening at Mount Sinai, Penn or Yale; any state newborn or congenital heart imaging registry; regional site lists for NEPTUNE, CureGN and TMIST; ReSPOND's current lead and members; HuBMAP and HTAN image formats from pages that rendered; imaging linkage in the MGB Biobank, Yale Generations and the Black Women's Health Study.

The instruments

Funding routes that reach an infrastructure purchase

Ranked by fit for a non-diagnostic software layer. None of these passes through a hospital IT capital committee.

Massachusetts · nonprofit AMCs and research institutes

MLSC Research Infrastructure

$1.5M – $5M
  • Funds equipment, hardware and software, service contracts and renovation over one to three years.
  • Current round open August 17 through November 6, 2026.
  • Repeat awardees: MGH ($3.379M Living Tissue Repository, Dec 2025), BIDMC ($1.807M MassNano), MIT (auto-stainer for a tissue and spatial imaging core, Jul 2026).
  • Over $638M invested to date.
Pennsylvania · NIH/NCI-funded institutions

CURE formula grants

~$36.5M/yr
  • Up to 50% may be spent on infrastructure, stated in the program's own annual report.
  • Allocated non-competitively by formula; no proposal contest.
  • SFY2021-22: Penn $6.56M, Pitt $6.56M, Penn State $5.58M, CHOP $5.58M. Current-cycle allocations are not published.
All states · NCI-designated centers

CCSG shared resource

$2.1M – $13.5M
  • Roswell Park's Aperio AT2 and image analysis platform are stated on its own page as funded by P30CA16056 and a C06 instrumentation award.
  • Most tractable at competing renewal. Live windows: Wistar's period ended 2/28/2026; Mount Sinai Tisch became Comprehensive 10/2025; Columbia in competing continuation; Wilmot newly designated 3/2025.
  • 19 NCI-designated centers across the region.
New York · medical schools

NYFIRST

Up to $1.0M
  • Capital equipment explicitly allowable, alongside lab establishment and upgrades.
  • Round 7 open 7/15/2026 through 7/16/2027, rolling review.
  • Requires a recruited or retained translational researcher and a 2:1 institutional match.
National · informatics tools

NCI ITCR

$300K – $600K/yr
  • Cancer informatics tools including imaging and pathology platforms. RFA-CA-27-019 U01 early-stage at $300K direct per year for three years; RFA-CA-27-020 and -021 U24 at $600K direct per year for five years.
  • Pre-application webinar September 15, 2026. Due dates not yet posted.
  • Institution-held, not vendor-held; a commercial layer participates as a subcontractor or in-kind partner.
Ohio · research institutions

Third Frontier TVSF Phase 1

$200K – $1.0M
  • Ohio research institutions are directly eligible. Institutional match required.
  • 300+ grants, ~$57M since 2012. FY2025 cycle: proposals due February, awards April.
  • Fits only where the purchase can be framed as validating an institution-owned technology.

Secondary routes, in order of size

  • Foundation and philanthropy. Dana-Farber took $779.2M in contributions in FY2024, Boston Children's $646.4M, Cleveland Clinic Foundation $459.2M, CHOP Foundation $163.7M, NewYork-Presbyterian Fund $186.1M. Historically 12.9% of hospital philanthropy goes to capital equipment (FY2012, the last public breakdown). Slow: six to eighteen months and a named donor attached to a clinical use case.
  • Internally adjudicated pools. Cleveland Clinic's VeloSano raised $14.5M in 2025 and distributes it by internal peer review. Nationwide Foundation's $10M Pediatric Innovation Fund (Oct 2025) names Biomedical Data Sciences, including cloud support for research and AI, as a pillar. MaineHealth's innovation program funds Ignite awards to $20K and Bonfire awards to $100K; external consulting is permitted, staff time is not.
  • CTSA hubs. Cincinnati Children's and University of Cincinnati renewed at $37.2M over seven years in Oct 2025, more than $65M with match, explicitly funding research infrastructure. Tufts CTSI $78.4M over seven years (2023); Rutgers-led NJ ACTS $47.5M; Yale $63.7M (2021); Buffalo $28.4M (2025).
  • Indirect cost recovery and residual funds. UVA's published policy returns 29.4% of indirect cost recovery to the clinical department that generated it; several institutions let departments keep the full residual on fixed-price awards. Mechanism is documented; magnitudes are not.
  • State AI compute. Empire AI (New York, more than $500M public and private, Mount Sinai and Rochester members), Massachusetts AI Hub ($31M state, $120M expected over five years, no hospital members named), Yale's $150M five-year AI commitment. These buy compute, not orchestration software, and are noted as signals of where AI budgets sit rather than as routes.
  • Departmental operating spend. 86% of healthcare IT executives report unsanctioned software purchasing in their system (symplr with CHIME, 2025). It exists and it is being actively consolidated (the share of systems running more than 50 applications fell from 60% to 51% over two years). Whether $30K clears at a director's signature at a given institution is unverified.
Gate map

What clears without a committee

Private nonprofit systems do not publish delegation-of-authority matrices. Public and university-owned systems do, and the ten states plus federal grant rules supply the rest. Everything below is quoted from a published policy or statute unless the row says [I], which marks the two rows that remain inference. Rows are the gates a $30,000 annual non-diagnostic software subscription passes through, on either the institutional or the grant-funded path.

GateWhat triggers itPublished thresholdsWhere $30K landsWho signs
Institutional purchasing — the path when the hospital or university pays
Capitalization line Whether the spend is capital or expense. A subscription with no perpetual license is expense. Commonly $5,000 [I]; no published hospital-specific benchmark found. Below it. Never enters the capital cycle. Nobody. This is why SaaS moves and perpetual licenses do not.
Departmental discretion Dollar value only. Ohio State: department may issue a PO under $5,000. Rutgers "Quick Order" up to $5,000. Temple: unit-level small dollar under $5,000. UVM: PurCard at or below $2,500. SUNY Upstate: detailed quote at or under $2,499.99. UConn: directors under $25,000. Above it nearly everywhere. Only UConn's $25,000 director band comes close, and $30K still clears it. Department head or administrator. Not reachable at $30K.
Purchasing office, quotes required Dollar value. This is the band $30K actually sits in. Penn State: $25,000–$99,999, "minimum three supplier bids required." SUNY Upstate: $2,500–$49,999.99 document price reasonableness, vendor responsibility profile from $15,000. UMass: $10,000–$50,000, two to three quotes, and the policy names "licenses (including software)." UVM: over $25,000 to $250,000, three quotes or a sole-source form. Ohio State: Purchasing approves at $5,000 or more. Squarely inside it at every institution with published numbers. Expect a quote exercise or a sole-source memo. Purchasing office, with a director-to-AVP signature. UVM: Director of Purchasing signs to $250,000. UConn: dean under $250,000, AVP under $500,000.
Formal bid or RFP Dollar value, and in two cases the word "software." Ohio State: no bid required under $75,000. Penn State: $100,000+. Rutgers: formal competition at $40,000 and above. SUNY: five sealed bids at $125,000 or more. UVM: RFP above $250,000. UMass: competitive procurement required at $10,000 per year. Erie County Medical Center: RFP over $20,000 for professional services and software. Below it at most, above it at three. ECMC, UMass and Rutgers would put a $30K subscription into competition. Purchasing, on a published solicitation.
Sole-source justification Buying without competition. Rutgers: waiver of bid required if the purchase is $50,000 or greater. UVM: single or sole source justification form in the $25,000–$250,000 band. UMass: no-bid justification. Ohio State: waived bids over $250,000 reported to the Board quarterly. Required where competition would otherwise apply. A vendor-neutral layer with no functional equivalent is the argument that gets written here. Requesting department drafts, Purchasing accepts.
State bid statute Applies to public and state-owned institutions regardless of internal policy. NY: discretionary to $150,000, OSC approval over $50,000, and $150,000 for healthcare facilities. OH: ~$79,600 for 2026 under ORC 9.17, and county hospitals inherit it via ORC 339.05. MA: three written quotes $10,000–$50,000. CT: open market to $50,000 with three quotes. RI: small purchase $10,000. ME: informal quotes $25,000 or less. NH: over $35,000 requires published review criteria. VT: simplified bid to $250,000, and AGO certification for services $25,000 or more per year. NJ: $100,000 for state colleges. PA: no thresholds apply to software license maintenance agreements and extensions. Under the ceiling in NY, OH, VT, NJ and PA. Over it in MA, CT, RI, ME and NH, where $30K triggers quotes or published criteria. Institutional purchasing, under statute.
Software pre-approval Category, not dollars. Some institutions gate all software regardless of price. Temple: "Software requests require pre-approval prior to TUmarketplace submission." Pennsylvania DGS exempts software license maintenance from thresholds entirely, in the opposite direction. Applies at any amount where it exists. IT, before Purchasing will process the requisition.
Value analysis or IT governance Characteristics, not dollars: PHI, LIS or EHR integration, clinical workflow change, net-new vendor, security review. Only one published charter with stated triggers was found. Covenant HealthCare Policy 641 sets capital review at over $5,000 and a new technology / high financial impact trigger at $500 or more per procedure or $25,000 additional expense per year, plus any product with less than 12 months of FDA clearance. Its new technology committee meets monthly and the policy allows 21 business days. $30,000 a year exceeds the one published financial trigger found. Non-diagnostic scope with no PHI and no EHR write-back is what avoids the characteristic triggers; the dollar trigger it does not avoid. Committee, meeting monthly or quarterly. Three to six months, up to a year.
Capital committee or board Capital classification, or a large dollar value. NYC Health + Hospitals: board approval at $10 million or more; micro-purchase to $100,000. UVM: board at $2 million, or $1,000,001 for professional services, or any contract over ten years. Ohio State: SVP approval for contracts over $1 million. Far below. Not a gate for this purchase. Board or trustees. 12 to 18 months on an annual cycle.
Grant-funded purchase — the path when a CCSG shared resource, CURE, MLSC or CTSA award pays
Classification under Uniform Guidance What kind of thing is being bought. This determines every gate after it. 2 CFR 200.1 defines intangible property to include "software, or software subscriptions or licenses." Equipment and supplies are both limited to tangible personal property, equipment being the lesser of the institution's capitalization level or $10,000. A SaaS subscription is intangible property, not equipment. It never enters the 200.313 equipment regime. Sponsored programs makes the call.
Micro-purchase threshold Dollar value against the federal MPT, which the institution may raise. Federal MPT rose to $15,000 on October 1, 2025 (FAR Case 2024-001). 2 CFR 200.320 lets an institution self-certify to $50,000, and higher with cognizant agency approval. Ohio State's approved threshold is $75,000, applied to federal and non-federal funds alike. Under the MPT at Ohio State and at any institution self-certified at $50,000. No quotes, no bid, no justification. This is the lightest gate anywhere in this table. PI and department administrator.
Simplified acquisition threshold Dollar value between the MPT and the SAT. Federal SAT rose to $350,000 on October 1, 2025. Above the MPT, 200.320 requires price or rate quotations "from an adequate number of qualified sources." Well below the SAT. Where the institution has not raised its MPT, $30K sits in the small-purchase band and needs quotes. Sponsored programs and central purchasing.
NIH prior approval Scope change or significant rebudgeting. Equipment over $25,000 is a scope-change indicator, but that applies to equipment and a subscription is not equipment. Rebudgeting is significant when a direct cost category moves 25 percent or more. Requests go to the grants management officer 30 days ahead and only the GMO's response is valid. Note the NIH Grants Policy Statement still quotes the superseded $250,000 SAT. No prior approval for the purchase itself. The realistic trigger is rebudgeting if the line was not in the approved budget. The authorized organizational representative signs, not the PI. The PI's signature is retained internally.
Cancer center governance Shared resource operation and access policy. The CCSG notice of funding opportunity requires applicants to describe shared resource "policies on operation and use" and charge-back systems, and requires "Center Director oversight of CCSG-supported Shared Resources." No NCI program officer concurrence requirement for a shared resource purchase was found, and no dollar threshold for one. Decided inside the center. Cancer center director and shared resource leadership.
The grant path is lighter than the institutional path at this price point. On the institutional side, $30,000 clears departmental discretion at every institution with published numbers and lands in the purchasing-office quote band, with three institutions putting it into formal competition and one published value analysis charter catching it on a $25,000 annual-expense trigger. On the grant side, the same $30,000 is a micro-purchase at Ohio State and at any institution self-certified at the 2 CFR 200.320 ceiling, which means no quotes, no bid and no justification. That is a difference in cycle time of months, and it is the strongest practical argument for entering through a funded shared resource rather than through hospital procurement.
Approach by status

What each status implies for a vendor-neutral layer

Deployed  Interoperate, do not displace

UPMC, Ohio State, NYU Langone, Mount Sinai, MSK, Cleveland Clinic, HNL Lab Medicine. The image management system is chosen and, at several, contractually locked for years. The open question at each is the archive and the cross-site or cross-modality movement of images, which the IMS vendor does not always own. Where the archive vendor is undisclosed, that is the first thing to establish.

  • UPMC's published state is roughly 5% prospective scanning at academic sites against a five-year goal of 100%; the multi-year ramp is where an orchestration layer would sit.
  • Mount Sinai, Ohio State and NYU are single-vendor Philips in pathology and, at OSU and NYU, Visage in radiology with the archive undisclosed.

Partial  Expansion is already funded or planned

University Hospitals Cleveland, Nationwide Children's, Columbia/NYP, MetroHealth, Yale, Brown University Health, Mass General Brigham. Live but narrow, and several have said publicly what comes next.

  • Brown University Health runs Philips, two Leica models and three Grundium units in one pathology informatics group; that is the clearest multi-vendor pathology scanner estate found in the region.
  • Nationwide Children's names incomplete LIMS integration, scanner failures and limited informatics bandwidth as its constraints, and is a cooperative-group biobank scanning at volume with an undisclosed platform.
  • Yale reads about 10% of caseload digitally with an unnamed scanner vendor and a Visage radiology contract whose 2017 seven-year term has run.

Intent  Procurement is live

Beth Israel Lahey Health, Penn Medicine, Rutgers Cancer Institute.

  • BILH ran a nine-month evaluation of nine image management platforms, has three finalists, and expects an eight-site launch in November 2026. In the same window it signed a seven-year A$90M Visage full-stack radiology contract with a Q1 2027 go-live and a migration off unnamed legacy PACS archives. Two migrations, two vendors, one year.
  • Penn's chair has named LIS harmonization across Penn Medicine entities as a goal; Penn is single-vendor Sectra in radiology and has already extended Sectra to genomics.

Greenfield  No evidence of a program

All of New Jersey, Northwell, Montefiore, Jefferson, Geisinger, most of New England outside Boston and New Haven, and most Ohio community systems.

  • For a non-diagnostic layer, greenfield in pathology is not by itself an opening; it means there are no images to orchestrate yet. The relevant greenfield accounts are those with a documented multi-vendor radiology estate or a live migration: WellSpan and Allegheny Health Network (Visage viewer over a separate VNA), Hackensack Meridian (single-vendor Fujifilm, network-wide Synapse Cardiology PACS 2025), Northwell (Nuvance integration, imaging plan undocumented).
  • Several are in financial distress and are noted as such in the table; those are deprioritized, not excluded.
Conditions

What changed recently

CMS and CAP diverge on computational pathology

In July 2026 CMS proposed removing computational pathology from the Clinical Laboratory Fee Schedule in both the OPPS and PFS rules, reclassifying ten algorithmic-analysis codes as "Software as a Medical Service" on the position that analysis of whole slide images does not require CLIA-regulated laboratory services. CAP opposes, citing checklist items GEN.50630 and GEN.52860. A non-diagnostic layer is not directly in scope, but every pathology informatics buyer is watching it.

Vendor consolidation

Tempus acquired Paige in September 2025 for $81.25M, with the ~7M MSK-licensed slides and Paige's Azure commitment. AstraZeneca acquired Modella AI, the Brigham Mahmood Lab spinout, in January 2026. Sectra completed its first US cloud migration (Emory) in June 2026. In KLAS's 2025 data Sectra is considered in nearly 60% of PACS decisions and most often selected; Philips, Optum/Change and GE customer bases are the ones most frequently evaluating options.

CAP inspects digital pathology directly

In 2025 CAP added a Laboratory General Checklist section, "Digital Pathology Including Remote Data Assessment": validation records, reports generated through digital analysis, scan-failure rates, image-quality criteria and QMS integration. The 2022 validation guideline still governs at 60 cases and greater than 95% concordance. This applies to diagnostic use and is one reason a non-diagnostic entry is faster.

Financial distress in the region

Jefferson −$196M FY2025 and −2.3% in H1 FY2026; UC Health $256M cumulative losses with 56 days cash; Yale New Haven −$196.8M FY2025 with no permanent CFO; Tufts −2.0%, fourth straight loss year; UMass Memorial −3.4%; Boston Medical Center −$240M with $129M from ex-Steward hospitals; UVM Health Network under a required $300M three-year cut; Mass General Brigham's largest layoffs in its history (Feb 2025); MSK −$47.9M with $177M of Epic go-live cost. Massachusetts' median hospital operating margin was −2% in FY2024. Against that: RWJBarnabas 6.4%, Cleveland Clinic 5.0%, MaineHealth 4.9%, ProMedica 8.1%, NYU Langone 3.1%.

Evidence gaps

What is not sourced

  • Radiology PACS and archive vendors for roughly half of the 69 institutions. Not found for Mount Sinai, Northwell, Montefiore, NYC H+H, Albany Med, all of New Jersey except Hackensack and Cooper, UPMC, Jefferson, Geisinger, Penn State Health, Main Line, Tower, Bon Secours Mercy, Summa, TriHealth, Kettering, Premier, UC Health, Dana-Farber, Boston Children's, BMC, Tufts, UMass, Baystate, Brown, Dartmouth, and most of Maine and Vermont. Job postings and KLAS decision records are the likeliest way to fill these.
  • Every departmental and service-line signature threshold in the gate map. No public benchmark for US hospitals exists.
  • Image infrastructure inside the cooperative-group biobanks (NRG, Alliance, COG) and inside CAIA. They scan or federate at volume and disclose neither scanner nor platform.
  • Image formats and hosting for HuBMAP, HTAN, SenNet, Kids First and CBTN histology. The portals list imaging as a data type; the pages that rendered do not say what format or which cloud. Award dollar amounts for most NIH Common Fund and NIA imaging programs are not on their public pages.
  • Mass General Brigham's current pathology vendor and primary-diagnosis status. The 2018 Philips partnership is documented; enterprise go-live is not.
  • AI dollar figures for MGB, Dana-Farber, Penn, Cleveland Clinic's IBM program, Northwell, Boston Children's, Dartmouth and Broad.
  • Current-cycle PA CURE allocations (public lists stop at SFY2021-22), CTSA amounts for most hubs, and the restricted share of hospital philanthropy (AHP member-paywalled).
  • Whether Case Comprehensive Cancer Center offers whole-slide scanning. Its shared-resources page does not say.
  • Connecticut Bioscience Innovation Fund's academic track terms (page unavailable), and state programs in RI, NH, VT.
  • Definitive Healthcare and Becker's bed counts do not reconcile; use one dataset per account.
Sources

Where the numbers came from

Becker's — health systems ranked by annual revenue, FY2025
GEN — Top 50 NIH-funded institutions, FY2025
ScienceDex — NIH RePORTER institutional totals
HHS TAGGS — P30 Cancer Center Support Grant obligations
NCI — Find a Cancer Center
Roswell Park — Pathology Network shared resource
Ohio State — Digital and Computational Pathology
Pitt/UPMC — Computational Pathology and Informatics
Laboratory Economics, Sept 2024 — UPMC per-slide cost
Executive War College — NYU Langone fully digital in one year
CAP Today, Aug 2026 — BILH digital pathology platform evaluation
Pro Medicus ASX filing, May 2026 — BILH Visage contract
Pro Medicus FY26 results — OSU, AHN renewals
KLAS, Aug 2025 — PACS market competitiveness
Brown University Health — Pathology Informatics scanner estate
Yale Medicine Magazine — digital pathology at Yale
NCI Imaging Data Commons — team and infrastructure
Nature Methods 2025 — HTAN data infrastructure
MSK — Cancer AI Alliance
NRG Oncology — biospecimen bank U24 report
Massachusetts Life Sciences Center — Research Infrastructure program
PA Department of Health — CURE grants
Empire State Development — NYFIRST
NCI — ITCR funding opportunities
Ohio Development — Third Frontier TVSF
NY Governor — Empire AI Beta online, Aug 2026
Yale — $150M AI commitment
Mass.gov — Massachusetts AI Hub
Pitt — Leidos $10M for CPACE
Cincinnati Children's — AI Imaging Research Center
Healthcare IT Today — Tempus acquires Paige
CAP — digital pathology CPT codes
Discoveries in Health Policy, Aug 2026 — CMS and CAP diverge
Becker's — Mass General Brigham FY2025
Becker's — Yale New Haven Health FY2025
Boston Globe — BMC and ex-Steward losses
symplr 2025 Compass Survey with CHIME
Advisory Board / Optum — capital spending trends
AHA Trustee Services — AHP philanthropy medians
philanthropy.org — Form 990 data
SUNY Upstate purchasing thresholds matrix, rev. Feb 2026
SUNY Procedure 7553, procurement, eff. Sep 2025
Ohio State purchasing policy
Ohio State, micro-purchase threshold raised to $75,000
Penn State bidding thresholds, Aug 2024
Rutgers procurement thresholds and bid waiver
UConn delegation of contract approval, Oct 2025
UMass competitive procurement, software licenses named
University of Vermont procurement policy
NYC Health + Hospitals procurement policy 100-05, Dec 2025
Erie County Medical Center procurement, RFP over $20,000 for software
Temple purchasing policy, software pre-approval
90 FR 41872, micro-purchase and simplified acquisition thresholds raised Oct 2025
2 CFR 200.1, software subscriptions as intangible property
2 CFR 200.320, procurement methods and self-certification
NIH Grants Policy Statement 8.1.2, prior approval
Covenant HealthCare policy 641, value analysis triggers
ORC 9.17, Ohio competitive bidding threshold
NY OSC, procurement thresholds
PA DGS procurement handbook, software license exemption
Children's Brain Tumor Network — metrics and Flywheel imaging holdings
AWS Open Data — ISIC Archive
KPMP — data and atlas repository
Penn CBICA — FeTS federated tumor segmentation
NYU Langone — fastMRI dataset
ADNI4 — site list
ABCD Study — sites and data sharing
NACC — SCAN neuroimaging
NACC — digital neuropathology pilot data stream
NIH Common Fund — HuBMAP funded research
HTAN — atlas centers
NIH Common Fund — SenNet funded research
TMIST — image transfer via ACR TRIAD (PubMed, 2023)
NCI — Quantitative Imaging Network
Breast Cancer Surveillance Consortium — registries
SEER Virtual Tissue Repository pilot (JNCI Monographs, 2024)
Rutgers registry-linked digital pathology framework (2022)
42 CFR Part 37 — NIOSH chest radiograph transfer requirements
World Trade Center Health Program — quarterly summary, Mar 2026
UPMC diabetic retinopathy program (PLOS One, 2024)
Framingham Heart Study Brain Aging Program — digital data
Mount Sinai AIR·MS — data modalities
Penn Medicine BioBank Vision Initiative (seminar listing, 2025)