AI as a Resource in Medical Assessment

Researchers and doctors at RISE, Karolinska Institutet, and Karolinska University Hospital are working together in a pilot project using artificial intelligence to find and interpret pathological changes in the adrenal gland that may indicate tumors and metastases.

The Challenge

Detecting tumors and metastases in medical images requires careful analysis by skilled radiologists. The process is time-consuming and the precision depends on human factors. With increasing demands on healthcare systems, there’s a need to support clinicians with advanced tools that can enhance detection accuracy.

The Approach

The team trains machine learning algorithms to identify and measure adrenal glands from CT scan images. The system learns to outline organ boundaries by analyzing hundreds of medical images, completing in minutes what typically takes significant manual effort.

The AI model has been developed through collaboration between RISE’s AI expertise and the clinical knowledge at Karolinska, ensuring that the technology addresses real clinical needs.

Key Finding

The results have been remarkable: “AI can identify the voxels of an organ better than the human hand.” The algorithm’s performance has improved significantly, with the computer sometimes outperforming radiologists in precision for specific measurement tasks.

Clinical Application

The technology aims to help radiologists:

• Detect adrenal tumors more accurately
• Calculate organ volume with high precision
• Streamline diagnostic workflows
• Support faster, personalized cancer treatment approaches

The Outcome

This project demonstrates how AI can augment rather than replace medical expertise. By handling precise measurement tasks, AI frees radiologists to focus on complex diagnostic decisions, ultimately supporting faster and more personalized cancer treatment pathways.