WP7 is focused on developing advanced AI tools to improve how PET/CT scans are analysed for patients with metastatic breast cancer. A PET/CT scan combines two types of images: a CT scan, which takes detailed pictures of the inside of the body (like a 3D X‑ray), and a PET scan, which shows how active different tissues are by using a tiny amount of a radioactive tracer. When these images are combined, they help doctors see both what something looks like and how it is behaving, making cancer easier to detect and monitor. WP7 aims to make scan interpretation faster, more accurate, and more consistent, allowing doctors to detect changes in the disease earlier. Early detection is important for adjusting treatment at the right time and improving patient outcomes.
A key part of WP7 is studying low‑dose PET/CT, which works the same way as a standard PET/CT scan but uses less radiation. This helps keep patients safer while still providing images that are clear enough for doctors to make confident decisions. WP7 is also developing AI algorithms that can track individual tumours over time and analyse differences within them, helping doctors better understand how the cancer is behaving and supporting more personalised treatment planning.
The AI tools will be carefully tested by clinical experts through both internal and semi‑external validation, without affecting patient care. These tools are meant to support, not replace, clinical judgement. Independent partners at other PREMIO COLLAB sites will also evaluate the tools to ensure they work well across different hospitals and scanner types. All development follows the EU AI Act and the Trustworthy AI Guidelines to ensure safety, reliability, and transparency.
Performance will be measured using well‑established evaluation methods, including accuracy, sensitivity, specificity, and how well the tools detect and outline tumours. The team will also check reproducibility (how consistently the tools perform across repeated scans or different centres) and robustness (how well the tools handle variations such as noise or different scanner settings).
WP7 also prioritises open science: all tools and algorithms will be made freely available to the research community. By combining innovative AI with a strong focus on patient safety and accessibility, WP7 aims to advance the next generation of cancer imaging—more precise, safer, and available to more patients.
None of the tools developed will be used for clinical decision making within the studied of PREMIO COLLAB. They will always be applied retrospectively. Recommended workflows for future use in clinical practice will always aim at ensuring control by the treating clinician.