AI outperforms experts in ovarian cancer ultrasound detection

Editor's Note

Researchers at Karolinska Institute in Sweden have developed artificial intelligence (AI) models that outperform human experts in detecting ovarian cancer using ultrasound images. As reported January 7 in DotMed, the AI was trained on over 17,000 images from 3,652 patients across 20 hospitals in eight countries, achieving an accuracy rate of 86.3% compared to 82.6% for human experts and 77.7% for less-experienced examiners.

The study highlights the potential of AI to address the global shortage of ultrasound specialists, DotMed reports, offering enhanced diagnostic support in complex cases and underserved areas. In a simulated triage setting, the AI reduced the need for expert referrals by 63% and lowered misdiagnosis rates by 18%, suggesting it could improve care efficiency and reduce costs for patients with ovarian lesions.

Researchers emphasized further studies are needed to examine the technology’s adaptability to various clinical contexts and patient populations. Although the research was conducted in collaboration with KTH Royal Institute of Technology and funded by Swedish research organizations, DotMed reports disclosure of potential conflicts of interest, including patents and ownership stakes in the diagnostic technology company Intelligyn.