The platform we developed for Radpoint had to cope with tasks such as patient registration, imaging (PACS), test description management or radiation dose control. The problem we faced was not only to ensure a seamless exchange of information between microservices, but also to create a system that was reliable, easy to maintain and fault-tolerant.

The multitude of services and the need to handle huge amounts of data required us to use a flexible and scalable architecture. What solution did we design and how did we move radiology into the cloud with Radpoint?

Multiple services on one platform ​

In designing the Radpoint platform, the biggest challenge proved to be the need to integrate many different services into a single system. In radiology, each service - from patient registration and image management (PACS) to study description and dose management - plays a key role, but operates under its own rules and requires specific data handling. To cope with this complexity, we had to design a distributed system that could not only handle each service individually, but also ensure a seamless exchange of information between them.

Challenges of moving radiology into the cloud ​

One of the biggest challenges we faced in the Radpoint project was the need to ensure data reliability and consistency in a distributed system that links multiple radiology services. Each of these services generates different types of data - from patient registration and imaging (PACS) to examination descriptions and radiation dose management. Differences in data structure, file formats and communication protocols greatly complicated the integration process.

We had to design mechanisms to ensure that all data was correctly recognised and processed across the platform. Even the slightest differences in data formats could have led to errors that were difficult to identify, which in radiology - where precision and accuracy are a priority - was a huge risk. In addition, the diversity of services meant that we had to design a system that was flexible, but at the same time resilient to errors and failures.

Using Apache Kafka to exchange information about events ​

To overcome the challenges of Radpoint's distributed system architecture, we chose to implement Apache Kafka as a key element of event information exchange between microservices. The distributed messaging system allowed for efficient and scalable communication between different modules of the platform, such as patient registration, PACS, test description management and radiation dose control.

Apache Kafka acts as a central hub through which all events generated by the system pass. Each microservice can publish data to selected topics, and other services can subscribe to this data and process it in real time.

This has resulted in a high level of data integrity and the ability to easily track all processes within the system. The solution provides not only scalability but also resilience to failures, which is crucial in a system handling critical medical data.

Apache Kafka also provides the ability to process large amounts of data with low latency. This allows the system to respond immediately to events, such as the introduction of a new study into the system or a change in patient status. With Apache Kafka, it has been possible to significantly simplify the management of inter-service communication and minimise the risk of errors due to data inconsistencies, which in the context of the radiology platform is of paramount importance.

What are the challenges of implementing a distributed architecture? ​

Although microservices offer many advantages, they are more complex to operate than a monolithic architecture. Each service requires separate monitoring, resource management and event handling. At Radpoint, this required the implementation of advanced monitoring tools, such as Apache Kafka, and meticulous management of inter-service communication.

Maintaining a distributed system can generate higher costs, both in terms of resources and management. Each service must be independently deployed, managed and scaled, which can increase operational costs.

Why did we decide to move radiology into the cloud? ​

• Scalability - each microservice can be scaled individually depending on the workload, allowing for efficient resource management. In the case of Radpoint, systems such as PACS or patient registration can run independently and their performance can be adapted to meet current needs.
• Independence - each microservice operates as a separate entity, which means that development, updates and maintenance of one service does not affect other parts of the system. This means that, for example, an update to the system for managing study descriptions can be carried out without disrupting other modules.
• Flexibility and adaptability - a distributed architecture makes it easy to introduce new features or services without interfering with existing components. In Radpoint, the addition of new functions related to radiation dose management, for example, can be done seamlessly without modifying the entire platform.
• Reducing the risk of platform-wide failure - the failure of one microservice does not affect the operation of other system modules. In Radpoint's case, if the module responsible for patient registration stops working, other key services, such as image management (PACS), will remain fully operational.

Cloud radiology implementation process ​

1. Systems and services analysis - it was crucial to identify the key radiology services that were to be moved to the cloud. We focused on functionalities such as patient registration, image management (PACS), examination descriptions and radiation dose control.
2. Designing microservices - we designed separate microservices for each service, which had to operate independently, but at the same time had to work together as part of the overall platform. Creating a modular structure allowed for greater flexibility and easier management of individual system components.
3. Use of Apache Kafka - we used Apache Kafka to ensure communication between microservices, which allowed for efficient exchange of event information in real time. This allowed any change in patient status or new diagnostic data to be processed immediately by the relevant services.
4. Iterative testing - the process of implementing microservices and their integration was done iteratively. We created unit and integration test suites for each service, which were progressively rolled out to ensure that communication between the systems was flawless.
5. Monitoring and data management - we implemented advanced monitoring mechanisms that allowed us to track traffic between microservices, log events and report potential errors.
6. Liaising with the support team - a key part of the process has been to maintain a close working relationship with Radpoint's support team and technology partners to quickly resolve issues as they arise and ensure high quality system performance.

This approach has resulted in a scalable and flexible cloud-based radiology system that effectively connects the various microservices to ensure efficient information exchange and uninterrupted operation in key areas of medical activity.