Faculty: A. Goshtasby (WSU) and M. Satter (WKNI)
Student: M. Jackowski
Sponsor: Wallace-Kettering Neurosciences Institute

The following demonstrates the steps of our segmentation method:

1. Select a volume of interest (VOI).

(a) Original image and selected VOI.	(b) VOI.

2. Select a seed point within VOI.

Interactively selected seed point.

3. Smooth VOI and find potential threshold values.

(a) Smoothed VOI.	(b) Histogram of smoothed VOI and 12 potential threshold values.

(a) Region obtained by thresholding smoothed VOI at 145.	(b) Region obtained by thresholding at 124.
(c) Region obtained by thresholding at 109.	(d) Region obtained by thresholding at 88.

4. Select the best threshold value interactively.

(a) Region obtained with best threshold value (109).	(b) Thresholded region after polygonization and shading.

5. Approximate region with a RaG surface.

(a)	(b)	(c)
(a-c) Rational Gaussian surfaces fitting to the region of interest with standard deviations equal to 0.01, 0.03 and 0.05, respectively.

6. Modify surface interactively to best fit image data (in progress).

The following images show examples of our segmentation process on other data sets.

(a) Original image.	(b) Thresholded region after polygonization and shading.

(a) Original image and selected VOI.	(b) Thresholded region after polygonization and shading.

This process can be used to segment a volume of interest or an entire image. The following show examples where entire images are segmented to extract the exterior skin of patients in MR images.

Published Preliminary Results

[WSU Home Page] [CSE Department Home Page] [Intelligent Systems Lab Page]

For more information contact: A. Goshtasby (ardeshir@cs.wright.edu).

Last modified: 7/23/99.