With the help of Artificial Intelligence, the accuracy and simplicity of identification of a rare disease such as cardiac sarcoidosis can be improved and streamlined, meaning that patients receive the care they need in a timely manner, and healthcare providers are able to use their resources effectively and efficiently. The Helsinki University Hospital (HUS) and DAIN Studios recently teamed up to implement an AI model to improve such a situation, and the results were very promising, paving the way for the future of AI in healthcare.

Business Value

We built an AI Deep Learning model to improve the overall diagnosis procedure for cardiac sarcoidosis with 93% accuracy – an accuracy that is as good as a highly-trained cardiologist.


Cardiac Sarcoidosis is a very difficult heart condition to diagnose. HUS cardiologists Dr. Jukka Lehtonen, Dr. Valtteri Uusitalo and their colleagues were engaged in what they referred to as the ‘old struggle.’ Some key areas of difficulty included:

  • Unusual metabolic activity in the heart is difficult for even the most experienced specialist to detect, using the best quality PET scans.
  • The only way of definitively knowing whether suspected cases are true or false alarms is by conducting a tricky needle-biopsy of the heart tissue.
  • Procedures often have to be repeated and can take weeks or even months to come to the correct conclusion.
explainable AI assists ti diagnose a rare heart disease at Helsinki University Hospital


The team had already been investigating ways to shorten the diagnostic time, looking at faster approaches based on RNA sequencing, with no success. A respected cardiologist at the hospital had been compiling a huge imaging data set and the hospital had over one thousand high-quality PET scans of suspected cardiac sarcoidosis cases.

When DAIN Studios came into the picture, it became clear that the use of this data would be unparalleled to anywhere else in the world, due to its sheer size. This gave the data scientists a huge dataset to work with which enabled for a more accurate training and implementation process to take place.

The goal was to train deep-learning algorithms to discern links between the patterns on the PET and CT images, and the eventual diagnosis based on laboratory examination of the biopsied tissue. This is the process that occurred:

  1. The team from DAIN first ensured that the data was anonymous in order to protect patient privacy.
  2. The images were pre-processed and the heart in both CT and PET scans were visually isolated.
  3. The data was split into training, validation and test sets.
  4. The preprocessed images were fed into a self-learning neural network, known as a residual neural network, which is popular for image processing.
  5. DAIN added a technique called attention-based modeling so that the software would not only tell users whether it computed an image to be a positive or negative case, but would also show them what patterns in the image made it decide this.
  6. Between June and November 2022, the DAIN team tested and tweaked the model so that it would catch as many of the test-sample’s cases identified by biopsy as possible.
explainable AI assists ti diagnose a rare heart disease at Helsinki University Hospital


Dr. Lehtonen from the Helsinki University Hospital predicts that if the software does as well with other datasets as it did with the Finnish one and then jumps through all the regulatory hoops, it could in three years’ time be in clinical use to at least rule out cardiac sarcoidosis in a diagnostic procedure.

For many people, it could potentially cut the diagnostic delay to days or weeks rather than the current weeks or months.” He does not see AI replacing the scientific certainty of the biopsy any time soon – although, of course, that is the ultimate goal. “The system is easy to use and understand,” he says. “We now need to gather experience with it, but it does look very promising.

Having been trained on an archive of combined PET/CT scans and diagnoses of hundreds of former HUS patients, DAIN’s software is currently over 93 percent accurate in identifying cardiac sarcoidosis – so-called true positive results – or its absence – so-called true negatives. “That’s as good as a highly-trained clinician,” says Ulla Kruhse-Lehtonen, one of DAIN Studios’ co-founders and CEO of the consultancy’s Finnish operations. “And with more data, it will start to do even better.” After training, validating and testing the software on a dataset from HUS, the next step will be to test it on data from elsewhere – talks with hospitals in the USA and Japan are underway.

This is the kind of ‘superpower AI’ that will make important contributions to our lives – if we use it correctly,” says Saara Hyvönen, another of the DAIN co-founders. “This AI does not decide for itself, it is an assistant to the doctor in charge, and it is what we call explainable AI – it ‘explains’ its diagnoses by showing what it sees or doesn’t see on the heat maps.Dirk Hofmann, a fellow DAIN co-founder and CEO of its German business, says: “Our vision is for this software to become part of medical-imaging machines, so that doctors not expert in cardiac sarcoidosis can still diagnose such a rare disease with confidence.


Jukka Lehtonen

Jukka Lehtonen

at Helsinki University Hospital

“The AI can see patterns in the images which the human eye cannot, this is of huge practical importance for medicine.” 


Helsinki University Hospital (HUS)

Helsinki University Hospital (HUS) is the largest provider of specialized healthcare in Finland, with 27,000 leading professionals treating almost 700,000 patients every year. They are a forward-thinking medical provider who are working together to protect and uplift the future of specialist medical care. Research into some of the most groundbreaking topics has led them to become acclaimed as one of the top healthcare providers. Their values of care, equality and pioneership are the foundations of what and how they provide to the community.


Data Analytics

The overall diagnosis procedure

Meet our experts

Hugo Gavert
Hugo Gävert
Chief Data and AI Officer
Ulla Kruhse-Lehtonen
Ulla Kruhse-Lehtonen
CEO of DAIN Studios Finland, Co-Founder