New book chapter at Trustworthy AI in Medical Imaging

We are excited to announce our new book chapter in Trustworthy AI in Medical Imaging. The chapter is titled “Collaborative evaluation for performance assessment of medical imaging applications” and is available here.

---

Abstract

With a growing number of medical imaging applications being developed for clinical use, there comes a need for extensive performance assessment to meet regulatory requirements as well as to build trust among healthcare stakeholders. Evaluation of these applications on external data can contribute to their objective and rigorous validation. Healthcare stakeholders can form partnerships to enable collaborative evaluation of medical imaging applications transparently and under neutral governance. Collaborative evaluation can unlock the power of diverse data, democratize the pursuit of clinically useful applications, and support regulatory compliance. This chapter offers background on validation in the context of product lifecycle, highlighting the importance of external real-world data; it introduces collaborative evaluation while discussing key considerations; it provides existing solutions that support collaborative evaluation; and finally, it lays out ideas for future improvements.