Multivision Diagnostics is developing a new AI tool for better cancer treatment. The tool provides a more precise diagnosis for each patient, a cancer fingerprint, allowing doctors to personalise treatment. This will improve both prognoses and quality of life.
At present, patients with tumours in the head and neck, for instance, may have to undergo major surgery because the course of their disease cannot be accurately predicted.
“If we had more precise diagnostics and found out that a cancer was not aggressive, we would be able to say that, for example, radiotherapy suffices,” says Research Director Sara Wickström of Multivision Diagnostics.
On the other hand, some patients do not benefit even from aggressive forms of cancer treatment. The tool developed by Multivision Diagnostics could help in establishing why some patients do not respond to treatment and finding new drug targets for them. This would enable the pharmaceutical industry to bring innovative cancer drugs to the market.
The software developed by Multivision Diagnostics creates a tumour-specific fingerprint by analysing data from patients’ tissue samples. First, a biopsy is taken from the tumour and stained with different antibodies to establish whether the cells contain biomarkers indicating the presence of cancer. This is done with the help of a technology that has been further developed by Teijo Pellinen, a researcher at the Finnish Institute of Molecular Medicine (FIMM).
Next, the AI software developed by Wickström’s team produces a computer image of the tissue structure based on the data collected. The structure can be modelled with the accuracy of individual cells to produce the patient’s cancer fingerprint. When it is compared with a database of information on patients with the same cancer, doctors can determine the patient’s prognosis and outcome, and select the best possible treatment.
“We now have at our disposal, for example, effective but expensive immunotherapies that don’t help everyone. With the cancer fingerprint, it may be possible to assess whether such therapies are likely to benefit an individual patient,” notes Wickström.
The researchers are testing their new technology on head, neck and colorectal cancers. Wickström believes that the tool will also be useful for diagnosing other cancers that form solid tumours and treating other diseases involving biopsies. In addition, the invention will reduce costs by saving pathologists’ time.
Alongside Wickström, the cross-disciplinary team behind Multivision Diagnostics includes biologists, physicists and bioengineers. The team has established how the structure of tissues controls cell function and how tissues change as a result of cancer.
At the core of the innovation is the insight that a more precise diagnosis can be obtained by looking for several different biomarkers in biopsies and examining tissue structures. At the beginning of 2023, the team received Research to Business funding to commercialise its technology. The project is coordinated by Postdoctoral Researcher Karolina Punovuori.
The timing is excellent, as digital pathology is expanding rapidly. The Nature Methods journal recently highlighted the integration of tissue structure analysis and marker analysis – the idea behind the Multivision Diagnostics tool – as its Method of the Year.
“We’re at the forefront of this innovation, which gives us a competitive advantage,” Wickström points out.
Mikael Muittari, the commercialisation specialist for the project, believes that innovations in digital pathology will increase dramatically in the coming years. The Multivision Diagnostics tool could be the first of its kind to be made available to doctors.
“No machine learning technology of this kind is currently in clinical use as a medical device. This can truly revolutionise cancer diagnostics,” he states.
The team is looking to foray into the European and US markets. The technology will be clinically validated for head and neck cancers by the end of 2024. With funding from Business Finland, the team will explore customer needs and develop the software. A patenting process has begun, and the aim is to establish a spinout at the beginning of 2025.
“We’re already in contact with Finnish private equity investors specialised in deep tech startups,” says Muittari.
Doctors do not have access to methods that would allow them to diagnose a number of cancers with sufficient precision or to predict the progress of the disease in each patient. This makes personalised cancer care difficult. Some patients undergo aggressive therapies unnecessarily, which reduces quality of life.
Multivision Diagnostics uses a computational tool to create a unique cancer fingerprint for each patient. Doctors can use it to assess the likelihood of the patient having an aggressive cancer and find effective treatments. The patient’s prognosis and quality of life improve.
Multivision Diagnostics has significant commercial potential among customers, including university hospitals and the pharmaceutical and biotechnology industry. Business Finland will fund the commercialisation of the innovation until the end of 2024, after which the aim is to establish an independent spinout.
The team is piloting the new technology in the diagnostics of head, neck and colorectal cancers. Next up are clinical validation and patenting. Multivision Diagnostics is seeking to collaborate with private equity and angel investors with experience in deep tech startups.