Digital pathology scanner procurement weakens when teams inherit a generic checklist and ask vendors to fill it in. A useful comparison begins elsewhere: with the exact diagnostic problem the system must solve. That decision then filters scanner architecture, regulatory posture, viewer and LIS relationships, and the workflow measurements worth collecting. Across the evidence base, no scanner emerges as universally superior. What differs is the fit between a scanner's documented behavior and a laboratory's slide mix, staffing, network, and intended use. A procurement team should therefore structure any evaluation around observable performance, not specification sheets. This article walks through that sequence, draws on published scanner comparisons and regulatory texts, and closes with a concrete worked example of how a single-institution tender might run a targeted head-to-head pilot without inventing a winner.
Start with the problem, not the product class
The College of American Pathologists frames acquisition around a single question: what problem is the institution trying to solve? [2] Primary diagnosis, frozen-section review, consultation sharing, and computational pathology impose different loads on a scanner and its surrounding systems. A scanner that performs well for archival slide digitization may be the wrong instrument for time-sensitive primary sign-out. CAP’s guidance points to a SWOT analysis as a way to separate real operational constraints from technology interest [2]. Without that step, no scoring model can be rational.
The problem definition also disciplines later vendor meetings. CAP warns that vendors present systems in the most positive light, and that absent clear use-case requirements, it is hard to judge fit [6]. A procurement team should write those requirements in operational language, not product vocabulary. For example, a requirement might specify “scan 250 mixed H&E, IHC, and special-stain slides between 5 p.m. and 7 a.m. with one operator,” rather than “40× equivalent brightness scanner.” That phrasing allows a vendor to respond with measurable behavior instead of marketing terminology.
Measured throughput matters more than vendor specs
Vendor throughput figures generally describe a theoretical 15×15 mm tissue area and omit the operator steps, slide variability, and failures that exist in daily practice [3]. A clinical comparison of sixteen high-throughput scanners from seven vendors illustrated how wide the gap can be. The authors recorded actual instrument run time from 7:30 to 43:02 hours for a standardized 347-slide set, while dedicated technician operation time ranged from 1:30 to 9:24 hours [3]. Total time, including operator labor, fell between 13:30 and 47:02 hours [3]. Those ranges are large enough to invalidate a procurement decision based on quoted scan speed alone.
CAP reinforces the distinction between capacity and speed. A high-throughput scanner can be continuously loaded or hold tens to hundreds of slides, but that does not guarantee fast scanning, because scan time depends on other factors [6]. Procurement teams should therefore ask each vendor to scan an agreed mixed slide set under local conditions. The comparison should log barcode reads, tissue detection failures, rescan events, and technician touch time. A scanner that finishes the batch slowly but unattended may be less costly in labor than a fast device that demands constant intervention.
A worked single-institution pilot, suggested as a plan
Consider a pathology department that wants primary diagnosis for breast, GI, and dermatopathology slides, roughly 500 slides per day across two shifts, with LIS barcoding and one technician per shift. The team would first reject any system unable to demonstrate 0.25 μm per pixel capture, continuous loading above 200 slides, and barcode compatibility with the existing LIS [1][5]. That preselection step aligns with ESDIP’s note that single-institution assessments often start with desktop information, conference demonstrations, and preliminary vendor discussions before targeted on-site testing [1].
The shortlisted systems would then scan an agreed 120-slide set composed of 60 H&E, 40 IHC, and 20 special stains drawn from the department’s recent cases. Technicians would record total instrument time, operator touch time, barcode failures, rescan events, and out-of-focus rates, using the artifact categories reported in the published scanner comparison [3]. The team should set decision thresholds in advance, such as an operator time budget and an error ceiling per use case.
This pilot would not produce a definitive national ranking. It would show whether a specific scanner meets the department’s documented requirements. The result would be a defensible procurement record tied to measured local performance, not a universal “best scanner” conclusion that the evidence cannot support.
Quality failures must be scored like throughput
Image quality cannot be judged by a single marketing image. The same 347-slide comparison found manual quality-control errors in 8% to 61% of digital slides per scanner run [3]. Specific artifact rates also varied sharply: missing tissue errors ranged from 0% to 21%, out-of-focus defects from 0% to 30.1%, and barcode failures from 0% to 26.2% [3]. Two scanners also produced tiling or overexposure artifacts [3]. These figures should caution any team against treating image quality as a pass/fail checkbox.
A procurement protocol should define failure classes in advance and set acceptable ceilings by use case. A consultation archive may tolerate occasional barcode errors resolved by manual naming, while primary diagnosis cannot. Regulatory benchmarks reinforce the same principle. FDA’s whole slide imaging controls require bench testing for color reproducibility, spatial resolution, focusing, tissue coverage, stitching error, and turnaround time, plus clinical non-inferiority data against optical microscopy [4]. Teams should ask whether vendor-supplied validation covers the specimen types and stains the laboratory actually receives, not a narrow manufacturer-selected set.
Resolution, optics, and scan modes determine the clinical ceiling
Scanner resolution choices should follow the tissue and application. An ESDIP guide states that 0.24–0.26 μm per pixel, usually described as 40× equivalent, suffices for routine histopathology, while 0.18–0.20 μm per pixel may be needed for cytology or specialized uses [1]. The same guide flags color fidelity as critical and expects calibration capability and consistent illumination [1]. A procurement team should not accept a single resolution across all requests. If cytology, hematopathology, or thick specimens are in scope, the lower-resolution option may be inadequate before the pilot begins.
Scan mode requirements are equally specific. Leica Biosystems asks whether the laboratory needs brightfield, fluorescence, or both, and whether live viewing for telepathology or multi-plane z-stacking for cytology is required [5]. Those capabilities are not interchangeable across all devices, and a vendor brochure may not disclose the absence of a mode unless asked directly. The ESDIP guidance also notes that compact bench-top systems holding 1–20 slides differ fundamentally from systems with capacities of 300–1,000 or more slides, with remote operation improving flexibility for after-hours scanning [1]. Matching those parameters to shift patterns is a procurement decision, not a technical afterthought.
Integration and regulatory posture create hidden total cost
A scanner’s utility depends on the systems around it. ESDIP advises verifying LIS compatibility and testing network bandwidth for multisite transfers, and it treats remote diagnostics and automated quality control as features supporting ongoing performance monitoring [1]. Leica’s guidance similarly asks whether the vendor offers off-the-shelf or custom LIS integration and whether the scanner can read the laboratory’s existing barcode format [5]. A scanner that requires manual file naming on every slide imposes a recurring labor cost that should be modeled over the device’s expected lifespan.
Regulatory choice interacts with integration. CAP notes that a closed FDA-approved system typically bundles a specified viewer, monitor, and GPU, whereas varying that setup creates an open-system approach that must be validated like a laboratory-developed test [6]. That is a consequential decision. A scanner buyer must also assess whether the intended tissue types and components fall within the FDA device indication [4]. Leica suggests a typical scanner lifespan of roughly five years and recommends asking about trade-in programs, service contracts, local support, and software patch frequency [5]. These support terms belong in the scored evaluation, because they determine whether a technically strong scanner remains operationally viable.
Sources & references
- Practical considerations for pathologists when selecting digital pathology systems: an ESDIP guide · boris-portal.unibe.ch
- DIGITAL PATHOLOGY 101 Packing Your Bags for a Digital ... · www.cap.org
- Digital slide scanning at scale: Comparison of whole slide imaging devices in a clinical setting - ScienceDirect · www.sciencedirect.com
- 21 CFR § 864.3700 Whole slide imaging system. · www.ecfr.gov
- Considerations for Buying a Digital Pathology Scanner · www.leicabiosystems.com
- Packing Your Bags for a Digital Pathology Journey · documents.cap.org
- Technical Performance Assessment of Digital Pathology Whole Slide Imaging Devices - Guidance for Industry and Food and Drug Administration Staff · www.fda.gov
- Digital slide scanning at scale: Comparison of whole slide imaging devices in a clinical setting - PMC · pmc.ncbi.nlm.nih.gov