INTRODUCTION
The Image Enhancement System (IES) has revolutionized the process of noise reduction in MRI over the last four years. IES uses patented GOP(r) technology, which does not smooth MR images (as most manufacturer's software does), but analyzes each pixel and eliminates much of the perceived noise within each image, leaving the anatomical structures sharp and pristine. The IES noise reduction system is particularly beneficial for mid- and low-field MR systems, where signal to noise is a major issue. However, high-field MRI systems also benefit from this technology, allowing these scanners to push the MR envelope (that is, use higher resolutions and faster scan times). IES also has allowed many older MR systems of all field strengths to remain in use and maintain their competitive image quality.
The patented GOP(r) technology used in the IES system incorporates advanced mathematical algorithms originally developed for image analysis and enhancement of military satellite data. This enhancement software was developed through a combination of image processing research and field experience over the past 15 years. To date, no other method has been as universally accepted in the demanding medical imaging market, where accuracy and clarity cannot be compromised.
LOW-PASS AND STATISTICAL FILTERING METHODS
The traditional method of reducing "noise" is to filter out high frequencies and let low frequencies pass through (a process known as low-pass filtering). However, since edges and lines in an image are predominantly high frequency, these structures will be blurred by the low-pass filtering process. This results in what is commonly referred to as "smoothing."
Another method of reducing noise is statistical filtering, which is a somewhat more sophisticated alternative to traditional noise filters. The basis of this method is that image areas containing structure and strong contrast between edges will have a higher variance in gray value than areas containing noise only. Therefore, in the statistical filtering process, the variance value of each image segment will determine whether noise reduction (low-pass filtering) is to be applied to that area. As a result of this process, noise is reduced and strong (sharp) edges are maintained in the image. Although an improvement over traditional low pass filters, statistical filtering methods have serious shortcomings. Weaker image structures, so important in MRI, often have the same variance as noise, which means that they will be filtered to the same extent as noise. Thus, while the statistical filters do somewhat take image content into account, these processes do not preserve the fine structures in the image. This is especially true of edges and borders that change gradually, such as small anatomical structures or subtle pathology.
THE IES APPROACH
IES uses GOP(r) methodology, which incorporates a revolutionary, patented two-step approach. This is a major technological advance over traditional and statistical filtering methods. In this method, the image is analyzed regarding the location of structures (edges and lines) as well as their orientation and curvature. Following this, the GOP(r) filtering process is performed. In image segment areas where no structures are found, noise reduction is automatically performed. Yet, unlike the low-pass and statistical filtering processes, the GOP(r) filter, through patented advanced image analysis, can distinguish image segments that contain both weak and strong structures, as well as detect the location, orientation, and curvature of these structures (including the weakest edges). This process determines the type of filtering to be performed and results in preservation of edges and lines in their original form, and, at the same time, enhances these important structures in the image. Therefore, the unique GOP(r) method can make an image sharper while simultaneously reducing noise.
Due to the computational complexity of GOP(r) methodology, the IES enhancement process requires a much more powerful computer than the typical MRI has. Therefore, the IES system has incorporated a specialized dedicated array processor and an advanced UNIX workstation to run the IES/GOP(r) software at 50 - 100 times faster than most standard MRI computers. Additionally, each IES system contains dedicated interface software and hardware that allow image data to be rapidly transferred digitally from the MRI system to the IES workstation for analysis and enhancement before returning the enhanced image to the host MRI computer for filming.
The combination of the IES interface, GOP(r) methodology, specialized array processor and UNIX workstation delivers image quality that has become the standard in the MRI industry.
BACK TO PREVIOUS PAGE