Skip to main content

Artificial intelligence for medical diagnostics

Johan Lundin is using mobile digital solutions and artificial intelligence to make diagnostics accessible, safe, and accurate in low-income countries. The tools he is developing can also reduce the workload of doctors and laboratory personnel in high-income countries.

Missing ALT text.
Professor Johan Lundin, credit: Ulf Sirborn.

What are you researching?

“I’m researching new ways of making diagnoses using digitisation and artificial intelligence, with particular focus on solutions for resource-limited countries. We are developing mobile microscopes and combining these with cloud-based artificial intelligence. In low-income countries, lab equipment is currently so expensive and professionals like pathologists are so few in number that many sick people never receive a diagnosis – or receive it too late. By making digital microscopes connected to the mobile network standard equipment in rural hospitals, and using AI to preview samples and reduce doctors’ workload, it will be possible to help many more people.”

How far have you gotten?

“We have made several prototypes of the microscopes, based primarily on components from the mobile phone industry. This is how we keep the costs down. The mobile industry’s high volumes provide low unit prices. We have also developed several AI algorithms that have learned to recognise different diseases in digitised samples. We are now testing the entire method in a screening study for cervical cancer in Kenya. We have also worked with AI diagnosis of parasitic diseases like malaria, schistosomiasis, and soil-transmitted infections, as well as AI classification of burn injuries. We have recently commenced a project involving the AI-assisted diagnosis of pneumonia.”

What are you hoping for in the long term?


“There are rapid developments within the fields of AI and mobile network performance, so I am
very positive about our ability to make the concept work. When the diagnostics for an initial disease are up and running, this can then be expanded to encompass additional diseases – essentially to all image- based diagnostics. Our research is also relevant for high-income countries: AI assistants that can
sort and preview medical images would also reduce the workload and shorten waiting times in Sweden. Imagine a scenario where we can give women their pap smear or mammography results immediately during their appointments, instead of through the mail several weeks later!”

Text: Anders Nilsson, in translation from Swedish
First published in From Cell to Society 2019

Johan Lundin

Professor of Medical Technology specialising in digital diagnostics at the Department of Public Health Sciences

Johan Lundin was born in Helsinki
in 1964. He graduated with a medical degree from the University of Helsinki in 1990 and received a doctorate from the same institution in 1996. Between 1997 and 2003 he was a postdoctoral researcher at the university’s Department of Oncology.

Johan Lundin also became an associate professor in biomedical informatics in 2003. He has conducted research at the Public Health Research Center in Helsinki, Helsinki University Central Hospital, and Institute for Molecular Medicine Finland (FIMM), where he has been its research director since 2011. He was guest professor at KI between 2012 and 2017.

Johan Lundin was appointed Professor of Medical Technology specialising in digital diagnostics at Karolinska Institutet on 9 January 2019.