A groundbreaking international study led by researchers at Sweden’s Karolinska Institute has demonstrated that artificial intelligence (AI) can outperform human experts in identifying ovarian cancer through ultrasound imaging. Published in Nature Medicine, the study offers promising implications for improving cancer diagnostics globally.
“Ovarian tumours are often discovered by chance, and there’s a critical shortage of ultrasound experts worldwide,” said Professor Elisabeth Epstein, senior consultant at Stockholm South General Hospital and lead researcher. “This shortage has led to unnecessary interventions and delayed cancer diagnoses. We aimed to explore whether AI could complement human expertise in this field.”
The research involved developing neural network models trained on over 17,000 ultrasound images from 3,652 patients across 20 hospitals in eight countries. These models were compared against human experts and less experienced ultrasound examiners.
The AI achieved an accuracy rate of 86.3 per cent, outperforming both experts (82.6 per cent ) and non-experts (77.7 per cent) in distinguishing between benign and malignant ovarian lesions.
In simulated triage scenarios, AI support reduced referrals to specialists by 63 per cent and misdiagnoses by 18 per cent, streamlining care and optimising hospital resources.
“This demonstrates the potential of neural network models to improve diagnostics, particularly in challenging cases and areas with limited access to expertise,” Professor Epstein added. Despite the promising outcomes, researchers caution that further studies are required to understand the full potential and limitations of AI in clinical settings. Current clinical trials at Stockholm South General Hospital aim to assess the safety and utility of the AI tool in real-world applications, with future research focusing on patient management and healthcare costs.
The study was conducted in collaboration with KTH Royal Institute of Technology and funded by several Swedish research organisations. Some authors, including Professor Epstein, have declared affiliations with Intelligyn, a company exploring AI-driven diagnostic tools.
_20250104114031_original_image_34.webp)
