Device pose estimation in fluoroscopy and DSA
This project focuses on recognizing guidewires, catheters and stents in interventional X-ray images. Colored landmarks and skeletons make device shape and movement visible across frames. The research connects device recognition with pose estimation and tracking in image-assisted vascular interventions.
Reading the demonstrations
The guidewire and catheter examples compare an original image with a recognition overlay. Straight-stent and bifurcated-stent examples compare an image overlay with the extracted structure. Each pair uses the same image crop and display scale so that the structure can be read against the image rather than being independently enlarged.
My contribution and research outputs
The portfolio presents my work on image-based device recognition and tracking, supported by the linked publications and patent records. These sources describe different research questions and contributions; they should be consulted for the scope of individual methods. Authorship and inventor position are listed separately on the research page.
What an animation cannot establish
A selected sequence illustrates output behavior but does not measure accuracy across patients, imaging angles or devices. Tracking quality requires independent test data, explicit definitions of landmarks or pose, and analysis of missed detections and false detections. The webpage does not claim clinical validation from the GIF examples.