MICCAI 2016 Daily - Tuesday

3D Reconstruction of Coronary Veins From a Single X-Ray Fluoroscopic Image MICCAI Daily: Where do you work, Maria? Maria: I am a postdoc at King’s College London. MICCAI Daily: What is your work about? Maria: I am working on guiding cardiac resynchronization therapy procedures. To do that, I reconstruct the coronary veins in 3d from a single X-ray fluoroscopic image. The way to do that is to first register the left ventricle onto the X-ray image which is contrast enhanced. MICCAI Daily: Why is it important? Maria: This is important because during cardiac resynchronization therapy, we place a lead into the left ventricle. We need to know how to pass the lead through the coronary sinus vessel through the left ventricle. MICCAI Daily: Why is that challenging? Maria: In X-ray fluoroscopy, which is basically the main method they use to guide the procedure, you can’t see the soft tissue of the heart. You can’t see the vessels very well. If you reconstruct them in 3d using MRI or something else, we can overlay that onto the X-ray image and help the clinicians see all of this information in 3d. What I want to do is reconstruct that in 3d, but instead of using MRI, to use just one X-ray image which is basically taken during the procedure. That is the main novelty of this work. MICCAI Daily: What algorithmic technique helped you to overcome this problem? Maria: First, I need to register the left ventricle onto one X-ray image. Then I use back-projection epipolar line reconstruction to reconstruct the coronary sinus in 3d just from one X-ray image. MICCAI Daily: What are the next steps of your work? Maria: The next step is to automate. To do that you need to annotate the coronary sinus into the X-ray image. The next step will be to use machine learning for the annotation. Everything will be done automatically. CVPR Daily: Thursday Presentation 17 MICCAI Daily: Tuesday Maria Panayiotou

RkJQdWJsaXNoZXIy NTc3NzU=