A radiology PACS is a software device that receives, transfers, displays, stores, and digitally processes medical images. The FDA defines it as a Class II device that provides one or more of those capabilities, and NIST describes it as a nearly universal authoritative repository for medical image information [3]. For a radiology department, the practical question is not whether to use a PACS but how to distinguish packages, validate interfaces, and structure an evaluation that produces a defensible decision. Primary documentation from GE HealthCare, Konica Minolta, Novarad, the VA, and a published procurement study shows that the useful distinctions are operational rather than purely technical. This guide explains those distinctions and identifies what the available documentation cannot establish.
Package Tiers and Deployment Choices
GE HealthCare describes True PACS as available in three versions matched to organizational objectives and budget. The technologist package is positioned for facilities that forward exams to external reading groups, the essentials package includes basic diagnostic reading, reporting, and workflow management, and the professional package adds native breast imaging workflow tools and EHR integration support [1]. These tiers are vendor-defined categories, not independent performance ratings. A facility that reads most studies internally would likely need at least the essentials package, while a site forwarding exams to a teleradiology group might evaluate the technologist package first. The documentation does not provide pricing or comparative throughput data for these tiers.
Deployment architecture also differs by product. True PACS uses a web-based architecture that GE HealthCare says supports modular, deconstructed, and distributed radiology deployments for remote reading and collaboration [1]. Novarad's NovaPACS Diagnostic Viewer includes web-based functionality, dual monitors, an administrative console, and on-site seven-year archives [7]. The Konica Minolta Exa PACS/RIS manual specifies minimum server and workstation requirements, including a Xeon Gold 5120 processor, 32 GB memory, and Windows Server 2022 for the server, with an Intel Core i5 or later and 8 GB RAM for user workstations [5]. These specifications are minimums for typical usage scenarios, and the manual directs users to contact a representative for recommended requirements for a specific implementation [5].
Interface Validation and Interoperability Evidence
The VA publishes a list of commercial PACS that have passed its internet validation test suite for HL7 communication with VistA Imaging. The December 2025 document lists products such as Agfa Impax 6.3.1, Carestream VuePACS 12.x, Fuji Synapse 3.2.X, GE Healthcare Centricity PACS 4.0 SP3, Merge PACS 6.5, Philips IntelliSpace 4.x, and Siemens Syngo Dynamics VA20 [4]. The VA states that this list does not constitute a product endorsement and that validation does not indicate how well a product will work operationally [4]. If a manufacturer revises any part of the PACS HL7 interface software, the interface must be re-validated [4].
The validation notes reveal functional differences that matter in procurement. Several products update patient location in real time, while Philips iSite 4.2 records patient location as it was at the time of the order [4]. Digisonics EC2-Software HL7 Manager has neither asserted support for nor been validated for report message processing and storage [4]. Volpara Patient Hub Dicom Export Service is described as an application-level service similar to scanning a letter into the PACS, not a modality sender that generates or transmits imaging data [4]. These distinctions are documented interface behaviors, not general quality judgments, and they show why a department should request validation evidence for the exact software versions it plans to run.
Hypothetical Evaluation Scenario for a Mid-Size Radiology Group
A mid-size group reading CT, MR, US, and CR/DX studies could adapt the published six-phase sequence as a suggestion, not a tested result. The team would first define pillars such as viewer speed, worklist behavior, interface validation, and archive retention. It would then request the VA-style HL7 validation evidence for the exact software versions under consideration, noting whether patient location updates in real time and whether report message processing is validated [4]. For remote reading, the team would compare True PACS's distributed deployment claims [1] against NovaPACS's web viewer and mobile device limitations [7]. For mammography, it would confirm that regulated monitors are available and that mobile devices are excluded from that workflow [1][7]. The team would then run anonymized local datasets through onsite demonstrations and score vendors on its own weighted pillars [2]. This scenario illustrates a decision sequence; it does not predict which product would win.
Diagnostic Viewing Constraints and User Responsibilities
Novarad's manual states that mobile devices are not intended to replace a full workstation and should be used only when there is no access to a workstation. Mobile devices are not to be used for mammography or fMRI, although they are used for diagnosis of images from CT, MR, US, CR/DX, NM, PT, and XA [7]. The manual supports diagnostic reading in lighting conditions from 7.3 lux in a typical reading room to 850 lux in a bright room, and it states that environments outside that range are not recommended [7]. The brightness setting on a mobile device should always be at 100 percent for maximum detail [7].
The Konica Minolta manual places responsibility for diagnostic quality on the clinical user. It states that measurement accuracy partially depends on image quality, which is subject to technologist skill, modality precision, and image resolution, and that users can help ensure diagnostic image quality and accurate measurements by using appropriate viewer settings such as window/level and zoom and by calibrating monitors [5]. The clinical user is responsible for judging measurement accuracy based on image quality and the accuracy of placed measurement points [5]. GE HealthCare notes that mammography images are still required to be reviewed on a regulated monitor [1]. These statements describe user obligations and device limitations, not vendor guarantees of diagnostic accuracy.
A Structured Procurement Sequence
A published open-access study in the Journal of Imaging Informatics in Medicine describes a six-phase PACS selection process: team formation, expectation setting, background assessment, initial vendor assessment, virtual demonstrations, and final evaluation [2]. The authors defined eleven pillars and 236 concepts, weighted them, and used surveys, standardized demonstrations, and a detailed request for proposal to score vendors on usability, integration, performance, and cost [2]. Five vendors were shortlisted after an initial presentation, three were selected for onsite assessments using department-generated anonymized datasets, and comprehensive RFPs and cost analyses were incorporated into final evaluations [2].
The study's outcome is specific to one department and should not be generalized. The committee recommended continuing with the incumbent vendor despite some identified deficiencies, advised that this decision be re-evaluated in three years, and the Radiologist-in-Chief accepted that recommendation [2]. The authors note that the optimal approach likely depends on system complexity, importance in day-to-day operations, the number of users and roles impacted, and cost [2]. This is a reproducible methodology, not evidence that incumbents generally win or that any particular vendor is superior. A department adapting this approach would need to define its own pillars and weights based on local workflows.
Security Scope and What Remains Unknown
NIST SP 1800-24 provides a standards-based reference design for securing PACS environments. The project considered viewer workstations inside and outside the healthcare delivery organization perimeter, workstations used by remote care specialists, emulated modalities, an emulated RIS, and PACS and VNA systems as systems of interest [3]. The practice guide is modular and can be deployed in whole or in part, and it includes installation and configuration instructions for products used in the NCCoE lab [3]. This reference design addresses security capabilities, not clinical performance or workflow efficiency.
Several important facts are not established by the supplied documentation. No source provides comparative pricing, customer counts, satisfaction ratings, or independent clinical outcome data for any PACS product. The VA validation list covers only HL7 communication with VistA Imaging and does not assess DICOM conformance, viewer ergonomics, or long-term archive reliability [4]. The GE HealthCare materials describe AI integration as optional and note that AI-based triaging and decision support depends on the regulatory cleared or approved intended use of each algorithm, and that it is not available with cloud/SaaS options in all regions [1]. A procurement team should treat these as open questions requiring vendor-specific evidence.
Sources & references
- True PACS | GE HealthCare (United States) · www.gehealthcare.com
- A Comprehensive Guide to Selecting and (Potentially) Replacing PACS: Navigating the Decision-Making Processes | Journal of Imaging Informatics in Medicine | Springer Nature Link · link.springer.com
- Securing Picture Archiving and Communication System (PACS): Cybersecurity for the Healthcare Sector · nvlpubs.nist.gov
- PACS System HL7 Interfaces Tested and Certified with VistA Imaging · www.va.gov
- Exa® PACS/RIS 1.4.32_P10 User's Manual · healthcare.konicaminolta.us
- SDD Fact Sheet | Picture Archiving And Communication Systems (PACS) · health.mil
- NovaPACS Diagnostic Viewer Manual November 2025 · info.novarad.net