Every day, thousands of MRI (magnetic resonance imaging) scans are taken from different patients. Up to now, radiologists have had to analyze each individual image manually when looking for tumors. Now artificial intelligence is helping to make the diagnosis: FUSE-AI, a start-up based in Hamburg, has developed a system that can detect and classify tumors on MRI scans. The system not only highlights abnormalities in the image, it even assesses whether the tumor is benign or malignant. This helps radiologists to detect significant tissue changes faster and more reliably. For the intelligent algorithm used to analyze the MRI images, the North-German start-up relies on IT resources from the Open Telekom Cloud. Features at a glance:
- FUSE-AI aims to improve cancer detection with artificial intelligence
- Self-developed algorithm analyzes MRI images, detects and highlights abnormalities such as carcinomas, and outputs an assessment as to its malignancy
- Leverages flexibly scalable, reliable and secure IT resources from the Open Telekom Cloud
The founders of the Hamburg-based start-up FUSE-AI already had a track record in the software development field. They developed apps for the medical sector. "Over time, it became increasingly clear that artificial intelligence would become an important tool in medicine," says Maximilian Waschka, one of the founders of FUSE-AI. "From this we had the idea to use machine learning, especially neural networks, to support physicians to make diagnoses." In 2015, four Hamburg entrepreneurs turned their idea into reality and designed initial versions of an algorithm that can detect symptoms of cancer automatically from tomography images. Their artificial intelligence was aimed at reliably finding, highlighting and classifying tumors.
"This is how we can contribute to the detection of many common diseases even more reliably and quickly. We not only reduce the effort and costs involved, but thanks to artificial intelligence we’re also increasing the patients’ chances of being cured.”
Sabrina Reimers-Kipping, PhD biochemist and co-founder of FUSE-AI
An MRI scan provides more than 2,000 images - per patient and per exam in the scanner. For a radiologist, the task is then to scrutinize the images for irregularities. They must also recognize and ignore so-called artifacts, or false alarms that may arise from errors in the imaging process. At the same time, they have to spot real carcinomas and ideally recognize whether the tumor is benign or malignant. A complex task that artificial intelligence can solve. Provided enormous computing power is available. But there is a limitation: “Our demand fluctuates. We only need very high capacities for the image analysis," says Dirk Schäfer, an expert in machine learning and co-founder of FUSE-AI. "Keeping this level of IT resources permanently available for such analyses would be very uneconomic." So the recent start-up was looking for flexible cloud resources to run the solution reliably and securely. "Another key advantage is the time we save. Because the platform services of the Open Telekom Cloud, such as the distributed Message Service (DMS), take on administrative tasks for us. This allows our developers to focus more on their core tasks," says Schäfer.
The Hamburg-based entrepreneurs applied to Telekom's start-up program: TechBoost. The ICT provider supports suitably qualified young companies with software-based business models with 100,000 euros of start-up funding for computing power from the Open Telekom Cloud, the public cloud service offered by Deutsche Telekom. In addition, TechBoost start-ups receive discounts on mobile, fixed-network and internet tariffs, and sales and marketing support. With their innovative solution, the founders of FUSE-AI qualified for TechBoost at the first attempt and are now making use of the scalable IT resources from the multi-certified, highly secure data centers of Deutsche Telekom in Saxony-Anhalt.
From 2018 onwards FUSE-AI will provide radiologists with round-the-clock diagnostic support from the cloud on demand. "The Open Telekom Cloud is a wonderful tool that provides us with the flexibility and scalability to implement a solution of this kind," says Schäfer. But that was just the beginning. FUSE-AI also intends to support physicians in other diagnostic imaging procedures. To achieve this, the founders, in conjunction with specialists such as dermatologists, have developed a dermatoscope, a device which takes pictures of the skin. Like the MRI images, these are also analyzed in the cloud using FUSE-AI’s artificial intelligence. "This is how we can contribute to detecting many common diseases even more reliably and quickly," says Sabrina Reimers-Kipping, a PhD biochemist and co-founder of FUSE-AI. “This not only reduces the effort and costs involved, but thanks to artificial intelligence it also increases the patients' chances of being cured.”