Vendor neutral artificial intelligence (VNAI) solutions such as the OSIMIS Platform offer greater flexibility and support for the clinical application of AI. The stated goal should be a long-term collaboration, based on a relationship of trust, good communication and a thorough understanding of the local workflow. A VNAI platform allows for a step-by-step approach that is more personalised, ensuring the evaluation and monitoring of the AI applications used are given the right amount of oversight and evaluation. Additionally, it is possible to roll out these applications in a broader context, such as in a network of different imaging centres, even if they have a different PACS.
Introduction
With the use of artificial intelligence (AI) now on the rise in the field of medical imaging, more and more hospitals have trialled or integrated one or more AI-based solutions into their diagnostic workflows. These figures will vary depending on whether the clinical practice is private, public, or academic.
For the implementation and integration of AI-based solutions, hospitals now have the following options:
- Integration by AI vendor directly: each AI vendor comes with their own integration stack.
- Integration through a PACS vendor platform: most picture archiving and communication system (PACS) vendors have now extended their traditional PACS storage and viewing capabilities with an AI offering.
- Integration through an independent AI integration platform: with the AI market now maturing, this kind of integration provider space has been on the rise for 2 years now.
Alongside these solutions, we are also seeing a more recent trend of individual AI vendors working together to offer their applications through one common platform. The initiative often comes from one of the vendors, usually the one with the largest installation base.
While option (1) enables AI vendors to work independently, it doesn’t respond to scalability & efficiency requirements that hospitals face in the long run. We will therefore leave it out of the scope of the current analysis.
The second and third possibilities (2, 3), are worthy of comparison. Here we take the OSIMIS Platform as an example of a vendor-neutral independent solution. As an objective complement to this document, we highly recommend you to read this article, which covers the key elements to consider when working with AI marketplaces.
Comparison
| |
OSIMIS Platform
|
PACS vendor platforms
|
|
Vendor selection
|
Osimis offers the possibility to search AI solutions from a very extensive repository of AI applications, while considering clinical evidence provided by the vendors. This is the result of a constant search for insights based on independent and objective criteria.
The OSIMIS Platform is open for integration with any vendor.
The department and/or the hospital have the final right of decision.
|
The PACS vendors have a restrictive approach towards a select number of AI vendors, based on criteria that are not necessarily in line with the needs of the hospital and/or the radiology department.
Moreover, implementing AI via the PACS vendor may lead to functionality limitations. The case becomes even more complicated in the case of migration to another PACS vendor or where several PACS systems exist within the same hospital network.
The PACS vendor determines what happens.
|
|
Scientific & clinical evidence
|
The OSIMIS Platform has a vendor-side portal, where AI vendors can upload all information about their products, including links to scientific evidence. The Osimis medical staff then assigns a clinical maturity score based on scientific results and provides the radiologists with a structured analysis.
|
Although it is difficult for Osimis to evaluate this aspect in particular, our conversations with several department heads have made us realise that obtaining scientific results and evidence is usually left to the radiologists themselves.
|
|
Pilots
|
Osimis can set up a secure environment to conduct retrospective pilots to evaluate different solutions, thereby ensuring that the PACS is protected from disruptive interactions.
|
Some PACS vendors have partial testing environment features available, not necessarily targeting the comparison of different AI-solutions
|
|
AI strategy
|
Osimis is a specialist in analysing the existing needs and in providing the most suitable solution.
As a pure AI integration platform, Osimis' core business is to assist in determining the right AI strategy
|
As AI is not the core business of the PACS vendor, it directly affects the ability of a PACS vendor to consider specific customer requirements.
|
|
Integration
|
DICOM & HL7 integration in PACS and RIS, that can be fully customized to the hospital’s needs.
It can also provide a unified vendor messaging & viewer.
In hospital networks, the OSIMIS Platform comes as a shared neutral interface despite heterogeneous PACS environments.
|
One of the key advantages of this approach is that seamless integration can be provided natively in the PACS.
However, implementing the AI solutions homogeneously across a hospital network holding different PACS will yield additional complexity.
|
|
Access to innovation
|
As the creator of Orthanc - the leading open source DICOM application for developers, researchers, and vendors of AI & medical devices - Osimis has a clear insight into innovation in academia and across the industry. This activity is closely monitored by our Chief Scientific Officer (Prof. Sébastien Jodogne) and Chief Medical Officer (Prof. Dr. Erik Ranschaert).
|
Unlike pure AI application providers, optimal AI integration is not a core focus area for PACS vendors. As a result, there may be gaps in communication between technical and medical teams, which in turn will lead to reduced and slower access to innovation in the industry.
|
|
In-house AI
|
Osimis has seamless integration of additional online applications and pipelines such as Robovision, which can be used to develop in-house AI projects based on its own data.
|
Some PACS vendors talk about it but as far as we know, none of the major PACS vendors can offer this kind of functionality.
|
|
Operational support
|
The Osimis Operations team deals specifically with AI integration. Osimis also has a Customer Success team that provides training and onboarding for users.
|
PACS vendors offer a general support or assistance service that redirects AI-related problems to an AI expert, often in a remote geography.
|
|
Monitoring & continuous improvement
|
The OSIMIS Platform provides a holistic dashboard and overview of the volumes of investigations processed by the various vendors, and the problems encountered.
Osimis’ Product team is also working towards an active communication & feedback loop solution for continuous evaluation of applications.
|
Although the technical support can reach the same level as what is offered by the OSIMIS Platform, the dependence on a PACS vendor could be a barrier.
|
|
Community
|
Experience sharing with the local and international radiological community.
|
Experience sharing only with radiology departments using the same PACS technology.
|
|
Pricing
|
Osimis facilitates volume-based cost savings at the network level.
In addition, the fact that Osimis is aware of the specific AI needs of the institutions and the specifics of the AI vendors allows for a better matching of supply and demand.
|
The pricing models are usually based on cost per study, which can lead to higher costs. In addition, there is no guaranteed possibility to use the volume model within hospital networks with different PACS systems.
There is a risk of overlapping applications, leading to license fees being paid multiple times.
|
Conclusion
In summary, vendor neutral artificial intelligence (VNAI) solutions such as the OSIMIS Platform offer greater flexibility and support for the clinical application of AI. The stated goal should be a long-term collaboration, based on a relationship of trust, good communication and a thorough understanding of the local workflow. A VNAI platform allows for a step-by-step approach that is more personalised, ensuring the evaluation and monitoring of the AI applications used are given the right amount of oversight and evaluation. Additionally, it is possible to roll out these applications in a broader context, such as in a network of different imaging centres, even if they have a different PACS.
Here are some key benefits for using the OSIMIS Platform:
- Vendor neutrality enables the radiology department and the hospital to have a customised AI policy that fully responds to their own vision and wishes regarding the application of AI solutions.
- The possibility to realise economies of scale by creating synergies at the hospital network level, even in a heterogeneous PACS setup.
- The personalised approach that facilitates an optimal integration in the workflow for all applications.
- The possibility to perform pilots on a smaller scale in a way that does not disturb the normal workflow and PACS activities.
- The ability to carry out a phased implementation of and investment in AI tools.
