A kidney is a dynamic space. Instruments push into the image, stones are grasped and disappear, and this is exactly what confuses navigation systems that are designed for static environments. So how can a ureteroscope still be located reliably in the renal pelvis? This is the question the MIRO Innovation Lab at DLR is working on, developing methods that specifically detect and classify moving objects in the surgical site.

Laura Oliva-Maza from the DLR Institute of Robotics and Mechatronics published a paper on this topic as lead author and presented it at the AECAI/CARE/OR 2.0 workshop held as part of MICCAI 2025 in Daejeon, South Korea. The extended version was published with open access in December 2025 in the journal Healthcare Technology Letters of the IET.

The paper is titled “Improving SLAM-Based Navigation in Flexible Ureteroscopy by Kidney Stone and Surgical Tool Segmentation” and builds on an existing navigation system of the institute that determines the position of the endoscope tip from the camera image. Until now, this system assumed a static scene. The team has extended it with image recognition that detects and segments instruments and kidney stones in real time: image points on instruments are discarded, while kidney stones are added to the map as potentially moving objects and marked in the visualization to facilitate the localization of the stones in the global map. The result: the position estimation of the ureteroscope improved by an average of 35 to 52 percent, depending on the method used. For urologists, this means more stable navigation during the procedure and a more accurate picture of where the stones are located in the three-dimensional model of the kidney.

MICCAI is one of the most important meetings worldwide for medical image processing and computer-assisted surgery. Every year, several thousand researchers, physicians and companies come together there. Around the three-day main programme, specialised workshops explore individual topics in greater depth. The AECAI/CARE/OR 2.0 workshop, at which the work was presented, is dedicated to endoscopy, augmented reality in the operating theatre and robotic assistance systems.

You can find the paper here.