3D Medical Ultrasound Image Segmentation
The goal of this research is to develop a robust automated system for the segmentation (boundary identification) of targets in 3D medical ultrasound images. Target structures include the prostate (cancer), free fluid volumes (such as abdominal bleeding), and cysts and lesions. Accurate and fast segmentation will allow doctors in the field to visualize 3D models of internal structures and fluid volumes as well as calculate statistics such as shape, volume, or track those statistics over time for a patient, giving the clinician additional diagnostic tools. The work is carried out by John David Quartararo, MS student in Electrical and Computer Engineering.
In addition to automatic segmentation, arbitrary 2D scan planes can be extracted from the 3D volume and viewed, providing views of internal structures otherwise impossible to acquire with traditional 2D B-mode ultrasound scans.
There are several approaches to image segmentation. Curvature evolution has been shown to be a powerful tool in image segmentation applications such as medical imaging and computer vision. The implicit geometric models (geometric/geodesic active contours using level set formulations) have seen increased focus in past years. In active contours, an initial curve (2D) or surface (3D) is subjected to several forces acting along the surface normal that control how it evolves through multiple iterations.
Test images were acquired by (i) simulating cysts in 3D using the Field-II software, (ii) building and scanning in 3D tissue-mimicking cyst phantoms with various level of contrast, and (iii) obtaining 3 sets of clinical 3D scans of prostate tumor. The 3D segmentation technique, using speckle reducing pre-processing, has been shown to provide a segmentation accuracy of the prostate on par with what experienced oncologists are able to achieve.
Maintained by webmaster@wpi.eduLast modified: March 11, 2008 14:05:58