Bilateral CNR-RS (UK)

Accurately measuring kidney size is vital for the diagnosis and monitoring of kidney diseases like Chronic Kidney Disease (CKD) and Autosomal Dominant Polycystic Kidney Disease (ADPKD). One of the most accurate ways of doing this is using Magnetic Resonance Imaging (MRI) scans to calculate the number of millilitres the kidneys take up. However, manually outlining the kidneys in these images is time-consuming and prone to human error. This collaborative project aims to develop an artificial intelligence (AI) system to automatically and reliably outline the kidneys in MRI scans. The goal is to work together to make this process faster and more consistent for use in real clinical settings. A key challenge with AI in medical imaging is that these systems can sometimes make mistakes if image quality is poor such as when a patient moves during a scan or when different MRI scanners produce slightly different results. To solve this, the project will also teach the AI to recognise when it might be wrong by estimating how confident it is in each result. We will test the system on real kidney MRI scans and also on images that we have deliberately added common problems to, like blurring or distortions to. This helps us make sure the AI works well even when things aren’t perfect, just like in real hospitals. The result will be a powerful new tool that gives doctors faster and more reliable information about kidney health, potentially improving diagnosis and monitoring for thousands of patients with kidney disease.