Early diagnosis of pancreatic cancer with AI - is the future already here?

Sammanfattning

Pancreatic cancer is one of the most aggressive and deadly forms of cancer. It has a very low survival rate primarily because of late diagnosis. This study aims to explore how artificial intelligence (AI) can contribute to improving the early detection of pancreatic cancer through the analysis of radiological images. The thesis is based on a qualitative approach that includes both an extensive literature review and in-depth interviews with three experienced radiologists. The findings from the literature study shows that AI technologies, specifically deep learning and radiomics, have the potential to detect pancreatic cancer several months before a clinical diagnosis is made. The interview study confirms that radiology remains as the primary diagnostic method in swedish healthcare, but that AI is currently only used as a tool for support in other areas. The participants had a positive perspective on AI’s role in enhancing early diagnostics and reducing diagnostic errors, while also emphasizing the need for robust datasets, clear legal frameworks, and ethical implementation. The conclusion is that AI holds strong potential as a complement to the work of radiologists, particularly by identifying subtle changes in imaging data that might otherwise go unnoticed. However, for the technology to be fully implemented, continued research, clearer distribution of responsibility, and increased acceptance among healthcare professionals and patients are required.