Boundary Detection

	.	CIMIL Home \| UROBOT \| CROBOT \| ART \| IMAGE \| GUEST BOOK \|
	.	Computer Integrated Medical Intervention Laboratory
Overview \| Image Archive \| Outlining \| Assessment \| Resources \|

Automated Prostate Boundary Identification

Wu Ruoyun, MEng student

Unlike manual operation, the extent of resection of robotic surgery is based on a cutting sequence generated by a computer program, which in turn relies on an image guidance system to map out the target organ, to obtain both its location and dimension. Automatically detect the prostate boundaries by means of computer algorithm is a fundamental step and is a mapping operation to enable a robotic intervention.

However, automatic identification of the contour in ultrasound images is more difficult compared to images taken from other modalities such as CT or microscopes. This is because of the poor contrast in intensity and inherent noise. In addition, the prostate interface with its surrounding tissue is not of a type where there is a sharp change in tissue density. This makes its identification even more difficult.

A 3-step procedure has been developed to overcome the above difficulties:


original image		Smoothed image		Edge Strength Detection		Boundary Recognition

The first step is 'Image Pre-process'. The purpose of this step is to reduce the noise as much as possible and preserve the edge information as much as possible at the same time. Gussian smooth filter is very effective for noisy ultrasound images.
The second step is 'Edge Strength Detection'. The purpose of this step is to calculate the edge strength for every pixel in the image. We developed an improved 'Radial Bas-relief' method to detect the edge strength (first derivative), which proved to be much more effective than its modalities.
The third step is 'Boundary Recognition'. A model-based adjustment combined with local search techniques was developed to recognize the prostate boundary from edge strength mapping images, which is the result of edge strength detector. A smooth, continuous, and single width ‘digital’ boundary line will be found in this step.