The first presentation in this session was by Tuncay on segmentation of the aortic valve area on CT. They have deviced a novel method to semi automatically segment the opening area of the aortic valve. They created an adaptive grayscale image to get the boundaries more clear and an adaptive algorithm to remove the calcium from the image. The technique proved effective and more reproducible in 21 patient datasets.
The second presentation by Haase was on Model based 3D registration of a CS catheter: application to single X-ray projections from a rotational angiographic sequence. They use a model of the catheter and combine this with the recorded Xray data to obtainna 3D representation of the orientation and movement of the catheter in 3D. They showed high accracy in the tracking in a simulated phantom setup. In clinical datasets also a high accuracy was obtained. Overall accuracy was 0.33 mm in 2D and 2.04 mm in 3D.
The third and final presentation by Masuda was on full automatic calculation of ejection fraction of left ventricle from either of short-axis view by processing succesive ultrasound images. The aim of their study was to design a method that does not require any initial input of the user. They envision to include this algorithm into a portable echography device for use in emergency situations. They demonstrate an interactive real-time detection of the LV cavity during the acquisition of the ultrasound. A subjective scoring by sonographers showed a high level of correct segmentation. Their algortihm was succesfull both in two and four chamber view. Good correlation was shown of the EF based on the new and conventional method.
The second presentation by Haase was on Model based 3D registration of a CS catheter: application to single X-ray projections from a rotational angiographic sequence. They use a model of the catheter and combine this with the recorded Xray data to obtainna 3D representation of the orientation and movement of the catheter in 3D. They showed high accracy in the tracking in a simulated phantom setup. In clinical datasets also a high accuracy was obtained. Overall accuracy was 0.33 mm in 2D and 2.04 mm in 3D.
The third and final presentation by Masuda was on full automatic calculation of ejection fraction of left ventricle from either of short-axis view by processing succesive ultrasound images. The aim of their study was to design a method that does not require any initial input of the user. They envision to include this algorithm into a portable echography device for use in emergency situations. They demonstrate an interactive real-time detection of the LV cavity during the acquisition of the ultrasound. A subjective scoring by sonographers showed a high level of correct segmentation. Their algortihm was succesfull both in two and four chamber view. Good correlation was shown of the EF based on the new and conventional method.
No comments:
Post a Comment