Incremental Volume Rendering Algorithm for Interactive 3D Ultrasound Imaging

Abstract

T1'ducuimelI. Lais bL-,-n approvedI for p btlic ieie(r sp ,and sale; its di&;trihul .on is-unlimited, The research reported herein was carried out with the partial support of NIH grant number POJCA4 7982.

Ultrasound Processing and Computing: Review and Future Directions

“…The only constraint to the clinician is to avoid moving the transducer too quickly during scanning, leaving gaps between 2D images. This can be minimized by providing real-time feedback to the clinician during scanning by displaying the volume incrementally as it is being scanned (74). To register the 2D images with freehand acquisition, a 6 degree-of-freedom (DOF) position and orientation measurement device can be attached to the ultrasound probe, which reports the location of the probe relative to a reference coordinate system.…”

“…To register the 2D images with freehand acquisition, a 6 degree-of-freedom (DOF) position and orientation measurement device can be attached to the ultrasound probe, which reports the location of the probe relative to a reference coordinate system. Both acoustical (65, 50) and electromagnetic (29,33,74) sensors have been used. These freehand systems are not as accurate as the mechanical systems, and they tend to be sensitive to environmental changes, such as temperature and humidity for acoustical sensors and electromagnetic interference for the electromagnetic sensors.…”

“…The most common reconstruction process involves distributing the pixel values of the acquired 2D images into a 3D Cartesian volume of voxels, which is an efficient format to later visualize (or render) the 3D data (74). With the position and orientation information, each image is spatially registered, transforming the coordinates of each pixel into a common reference coordinate system of the volume.…”

“…linear, fan, and rotational) that have a predetermined geometry with a well-defined scanning motion. The unconstrained motion for freehand systems is better suited for forward mapping, in which the input pixels are mapped to locations within the reconstruction grid and their pixel values distributed to the nearest voxels in the volume (74). Simple and fast techniques to distribute the pixels are widely used, such as replacement value, maximum value, and pixel averaging (33).…”

“…Many researchers have evaluated 3D ultrasound systems trying to meet the computational requirements. Ohbuchi et al (74) developed an interactive 3D system with a workstation to digitize sequences of 2D images from a commercial ultrasound machine, a mechanical tracking arm (3 DOF), and a powerful computing platform to reconstruct and render the volume. For an acquired sequence of 256 ‫ן‬ 256 images, they reconstructed a 128 ‫ן‬ 128 ‫ן‬ 128 volume at 1.7 fps and rendered a small volume (65 ‫ן‬ 81 ‫ן‬ 100) by using a special graphics engine (Pixel-Planes 5) at 10 fps.…”