Industry Standards Hold The Key to AI Adoption in Enterprise Imaging

AI applications are entering radiology at rapid and increasing rates. AI distinguishes patterns and irregularities in large collections of data, which makes radiology an ideal application. More than other medical disciplines, radiology has a long history of storing studies digitally, with plenty of images to train AI algorithms. Algorithms draw from millions of digital images and can aid diagnostics. As an example, the CINA® Head triage AI solution from our partners at Avicenna.AI enables rapid detection of vascular events that could signal an impending stroke.
Despite its promising future, AI faces a number of obstacles in enterprise imaging. In a previous article, we reviewed the challenges of deploying AI into live clinical production, as observed from Dicom Systems’ position as a trusted Health IT partner to top healthcare facilities. The bottlenecks that hinder successful AI implementations include inconsistent application of imaging industry standards and APIs, the prevalent reliance on cloud solutions in order to use an algorithm, insufficient imaging workflow knowledge on the part of AI vendors, and inconsistent AI results formats and delivery.

Imaging Industry Standards Enable AI Adoption

We have found that overcoming the first obstacle—inconsistent application of imaging industry standards and APIs—is quite surmountable in today’s AI development environment. While AI developers may not be “native speakers” when it comes to enterprise imaging standards, they are often capable of adopting industry standards within their processes.
Industry Standards Hold The Key to AI Adoption in Enterprise Imaging
Industry standards, such as DICOM, DICOMweb, HL7 and FHIR are paramount to ensuring that AI can be safely and effectively deployed in production clinical environments. RESTful APIs (DICOMweb and FHIR) in particular offer much by way of interoperability. The following 7 basic elements, combined within a cohesive platform, can enable enterprise imaging environments to adopt new AI technology without disrupting the existing infrastructure or clinical workflows.


Digital Imaging and Communications in Medicine (DICOM) is the standard for the communication and management of medical imaging information and related data.


DICOMweb™ is the DICOM Standard for web-based medical imaging. It is a set of RESTful services, enabling web developers to unlock the power of healthcare images using industry-standard toolsets.


HL7 is a set of international standards for transfer of clinical and administrative data between software applications used by various healthcare providers. HL7 standards support clinical practice and the management, delivery, and evaluation of health services.


Fast Healthcare Interoperability Resources (FHIR, pronounced “fire”) is a standard describing data formats and elements (known as “resources”) and an application programming interface (API) for exchanging electronic health records (EHR).
FHIR builds on previous data format standards from HL7, and is easier to implement because it uses a modern web-based suite of API technology, including a HTTP-based RESTful protocol.


A docker is a tool that helps users to leverage operating-system-level virtualization to develop and deliver software in packages called containers. Containers are a standard way to package an application and all its dependencies so that the application can be moved between environments and run without changes.
Docker has provided a set of tools to simplify the use of containers. This has led to rapid adoption of containers by developers and operators. By connecting via standard Docker, the AI router is able to integrate ‘n’ numbers of AI algorithms that can interchangeably reside on premises within the caregiver’s IT enterprise, or in a cloud infrastructure (public, private or hybrid).

TLS Encryption

Transport Layer Security (TLS) is a cryptographic protocol designed to provide communications security over a computer network. Websites can use TLS to secure all communications between their servers and web browsers. When it comes to enterprise imaging, security is of critical importance. The DCMSYS Unifier appliance supports DICOM 3.0 standard TLS encryption.

How Dicom Systems AI Conductor Enables AI Adoption

At Dicom Systems, our technology has always been designed to function as a “universal adapter” for enterprise imaging, with interoperability as the guiding principle. To achieve this vision of nimble interoperability, the AI Conductor was designed as a Linux-base appliance, managed via a simple browser environment. By leveraging the industry standards discussed above, our platform integrates seamlessly with hospital information systems, including PACS, RIS and EHR.
In a 6-step process, images and other health information flow through the AI router platform as illustrated below:
  1. The AI router receives DICOM images or other imaging-related objects and scans all metadata as they are coming in
  2. AI router flags any predetermined metadata signaling that a set of imaging objects are candidates for one or several AI algorithms, some of which may be on prem, and others in a cloud. For cloud-based algorithms, the AI router platform can be programmed to automatically de-identify pixel and metadata in order to maintain strict HIPAA compliance.
  3. The AI router signals to the appropriate docker that a study has been identified for further processing by the algorithm
  4. The AI router pushes metadata via QIDO-RS and pixel data via WADO-RS to the appropriate docker
  5. The docker analyzes the data and returns a result back to the AI router via STOW-RS
  6. The AI router receives the results from the docker, translates the result into the appropriate standard expected by the end-user system: DICOM-SR to PACS and/or dictation system, HL7 or FHIR to EHR/EMR, DICOM overlays to the PACS

What’s Next In AI Adoption

Enterprise imaging workflows, with or without AI, already demand substantial scalability, as data sets continue to grow in size and depth of complexity. As the imaging community embraces and integrates AI, a few factors will continue to influence the rate and success of adoption.

The classic elements of IT performance will continue to play an important role in replicability, scalability and performance. As Executive Vice President Florent Saint-Clair said, “We cannot outrun the laws of physics.” For AI algorithms to perform at their fullest potential, the IT infrastructure itself must be up to the correct standards. Storage speeds (I/Ops), available bandwidth, CPU power, and memory continue to be critical factors in how AI solutions can consistently perform under varying IT conditions.

AI vendors must collaborate with the enterprise imaging industry to deploy their algorithms in a way that does not impair other software within the ecosystem from functioning at a level of efficiency that has become the accepted norm in enterprise imaging. The demands of imaging workflows, augmented by the time-sensitive nature of diagnostics, point to industry standards as the surest way to address the performance and scalability expectations of imaging enterprises. A dockerization of AI will ensure the replicability of these processes.

We are excited to be part of the AI enablement journey in enterprise imaging. To learn more about AI Conductor, we invite you to download our white paper, or schedule a demo with our team.