AI in radiology: a platform simplifies clinical workflows

12.05.2026

Artificial intelligence is about to fundamentally change the healthcare sector. But what does its use in everyday clinical practice actually mean? This is a question that the Munich-based health tech company deepc is also addressing. With deepcOS, the scale-up has developed a cloud-based, manufacturer-independent platform for radiology that makes it possible to seamlessly integrate a variety of AI-supported imaging and diagnostic tools into existing workflows. The result: optimized workflows, precise diagnoses and greater patient safety. In an interview with Bayern Innovativ, co-founder and CTO Dr. Julia Moosbauer explains how this works, what hurdles still exist - and why AI is no longer just about algorithms.

Ms. Moosbauer, how would you describe your company's mission in a few words to readers who are not yet familiar with deepc?

Julia Moosbauer: At deepc, we are developing a technological infrastructure that enables hospitals to use AI in radiology in a secure, simple and scalable way. Our goal is to enable healthcare providers to easily integrate, manage and scale AI solutions into everyday clinical practice. In doing so, we are creating the technological basis for transforming clinical workflows in hospitals in the long term and transferring AI from an individual tool to a genuine infrastructure for daily use.

AI is often discussed in very abstract terms. What concrete changes can it bring to everyday radiology?

Julia Moosbauer: That depends very much on the specific use case. For example, AI can help radiologists to prioritize their work. By evaluating image data from examinations in advance with AI and highlighting abnormalities, they can assess more quickly which cases should be given priority. In some cases, AI also supports the writing of reports or even pre-fills them. All in all, AI can therefore provide support at many different points in everyday clinical practice.

Can you illustrate this with a tangible example?

Julia Moosbauer: Imagine a young junior doctor who is on duty alone at night and the emergency room is overcrowded. What should he look at first? AI can help them to prioritize cases. It can provide a second opinion, for example by specifically looking for fractures, so that even subtle findings are not overlooked in stressful situations. AI can also support him in writing the findings by preparing parts of them in advance. I think this is a very tangible example of the benefits of AI in everyday clinical practice.

You are relying on a platform with access to various AI tools instead of a single model. Why is this the better approach for hospitals?

Julia Moosbauer: Hospitals work with various imaging procedures. There are X-rays, CT scans and MRI examinations for many different use cases and body regions. That's why we need different models for different applications. The market is also developing extremely quickly. A model that is leading today may already be outdated in one or two years.

Our platform is deliberately designed to be neutral. It makes it possible to evaluate, compare, integrate and monitor different AI models. And all with a single installation. This gives clinics the flexibility to always use the best available technology.

In your opinion, what are the biggest hurdles that hospitals face when they want to integrate AI into everyday clinical practice?

Julia Moosbauer: There are several hurdles that hospitals have to overcome.

Firstly, regulation. Strict requirements apply in the healthcare sector, for example with regard to data protection. The AI Act will make this even more complex.

Secondly, integration into existing IT systems. Radiologists don't want any additional clicks or detours, which means AI has to fit seamlessly into the workflow.

Thirdly: the human factor. At the beginning, you have to sit around a table together and define clearly: What problem do we actually want to solve? It's not about simply applying AI to everything, but about understanding the specific clinical challenges together with the users. Only then can we build trust and really involve the people who will later work with the technology.

Our task is not only to provide them with technological support, but also to overcome these hurdles together with them.

Trust is a key issue in medical AI. How do you ensure that AI supports doctors without them relying on it too much?

Julia Moosbauer: That is a very important point. Monitoring after the introduction plays a decisive role here. AI must be continuously monitored in clinical use, be it in terms of usage, performance, possible biases or changes over time.

Excessive trust in AI, the so-called "automation bias", is a key aspect here. After all, it is not just about the performance of an algorithm, but also about the interaction between humans and AI in everyday clinical practice. Training users in how to use the technology is therefore an important aspect. They need to understand how AI systems provide support, where their limits lie and how results should be classified correctly. At the same time, we are working on continuously creating further measurable criteria that will help us to check whether AI is being used safely and responsibly in everyday clinical practice.

What misconceptions do you think exist in the general public when it comes to AI in medical imaging?

Julia Moosbauer: Many people think that progress primarily means better algorithms. In reality, the bigger challenge is to integrate AI into clinical workflows, ensure smooth collaboration with healthcare professionals and monitor AI effectively.

It's not just about the performance of a model, but about providing the right information at the right time in the right system to support human decision-making.

Is your technology already in use?

Julia Moosbauer: Yes, our solution is being used in numerous clinics in Germany, the UK and the USA. Our users, mainly hospitals and university clinics, are actively working with several AI tools via our platform.

What could be real advances for radiology in the coming years?

Julia Moosbauer: We are currently experiencing the transition from individual pilot projects to AI as an integral part of everyday clinical practice.

In my view, progress means achieving measurable improvements, such as shorter waiting times, faster diagnoses or earlier detection of diseases, for example in early cancer detection. At the same time, AI can help to mitigate structural problems such as staff shortages or cost pressure. AI offers enormous potential here.

Thank you very much for the interview, Ms. Moosbauer!

deepc GmbH is a partner in Bayern Innovativ's Health Partner Network. As an innovation agency, Bayern Innovativ brings together players from science, industry and practice, creating spaces for exchange and knowledge transfer. On behalf of the Free State of Bavaria, it supports companies and start-ups in gaining orientation in the complex innovation ecosystem. In this way, promising approaches are sustainably transformed into marketable solutions and Bavaria is strengthened as an innovation region.