AI Tools Slowly Gain Traction in Medical Imaging, But Full Automation Remains Distant

Early predictions that AI would replace radiologists have not come to pass, but AI tools are increasingly used in medical imaging.

Current AI tools serve specific functions like flagging abnormalities, but have limitations like focusing on common issues rather than full diagnostic assessment.

Researchers are exploring "foundation models" trained on huge datasets that can be adapted for diverse tasks, with potential to act more like a physician.

Tech companies are developing multimodal models combining images, text, genomics data, etc. aiming to improve on current tools' capabilities.

Experts say while capabilities are advancing, there is a long way to go to demonstrate these new AI tools can be safely implemented in clinical care.