AI improves ovarian cancer diagnosis

Photo Credit: Shutterstock

Researchers at the Karolinska institutet in Sweden have conducted a new study that shows AI-based models are superior to human experts in identifying ovarian carcinoma in ultrasound images. This study was published by Nature Medicine. Professor Elisabeth Epstein, senior consultant in the Department of Obstetrics and Gynecology at Karolinska institutet and Department of Clinical Science and Education at Sodersjukhuset at Stockholm South General Hospital says

“Ovarian tumours are common and are often detected by chance,” . AI performs better than experts

Researchers have created and validated neural networks that can differentiate between benign and cancerous ovarian lesions. They trained the AI using over 17,000 images of 3,652 patients from 20 hospitals across eight countries. The researchers compared their models with the diagnostic capabilities of a group of ultrasound experts as well as less experienced examiners.

Results showed that AI models were more accurate than both experts and nonexperts in identifying ovarian carcinoma, with an accuracy rate 86.3 percent, as compared to 82.6% and 76.7 percent for expert and nonexpert examiners, respectively.

“This suggests that neural network models can offer valuable support in the diagnosis of ovarian cancer, especially in difficult-to-diagnose cases and in settings where there’s a shortage of ultrasound experts,” says Professor Epstein. Reduce the need to refer experts

 

AI models are also able to reduce the number of expert referrals. The AI-supported triage reduced the referral rate and misdiagnosis rates by 18% in a simulation. It can result in a faster, more efficient and cost-effective treatment for patients who have ovarian cancer.

The researchers emphasize that despite the encouraging results, further research is needed to fully understand the potential and limitations of neural network models. Filip Christiansen is a doctoral candidate in Professor Epstein’s research group, at Karolinska Institutet. He also shares the title of first author of this article with EmirKonuk from KTH Royal Institute of Technology. Evaluation of AI Support Safety

Researchers are currently conducting clinical trials at Sodersjukhuset in order to assess the AI Tool’s everyday safety. A randomised, multicentre trial will be conducted in the future to assess its impact on healthcare costs and patient management. The research was carried out in collaboration with KTH researchers and funded by the Swedish Research Council, Swedish Cancer Society, Stockholm Regional Council, Cancer Research Funds of Radiumhemmet and Wallenberg AI, Autonomous Systems and Software Program.

View Article Source

Share Article
Facebook
LinkedIn
X
3 months until cheaper olive oil hits supermarket shelves
Mondelez International warns of a price increase for cocoa
Mental Health Problems? It Might be Better In The Morning