 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
What is Signature Mapping™?
Signature Mapping™ is the next step in technology evolution beyond pattern recognition. It is a dynamic, iterative process, in which specifically designed algorithms impact image pixels causing the group of pixels representing each tissue to respond in a unique collective way.
This response-based reaction generates new groups of “self-classifying” pixels. These self-classifying groups provide a unique set of signatures for each tissue or material. The resulting “signature mapped” data, enables the implementation of a rich set of imaging tools that provide new clinical approaches for medicine. These include image clarification and visualization, feature extraction, tissue characterization and quantification, disease detection, monitoring and staging. Process works effectively across all imaging modalities.Signature Mapping™ applies a dynamic and interactive process, in which unique Guardian algorithms impact image pixels in ways that cause groups of pixels related to one another through their association as part of the representation of a particular material (e.g., a lesion) to react or respond collectively.
Then, through a repetitive, iterative application of the algorithms, such associated pixel groups are segmented into image objects which continue to appear more distinctly related to one another – regardless of their distribution within the image – while simultaneously appearing more distinctly different from the pixels of other materials. Ultimately, such distinctions emerge as the unique signature for that material.
In image processing terms, Guardian’s mathematical algorithms work together in a processing (the “pixel impact/reaction” part), and the other (the “repetitive, iterative” part) applies a series of polynomial transformations called Iterative Transformational Divergence, or ITD. The result is object segmentation and differentiation.
With each run of the ITD process on created new image data, a new hyperplane is created containing additional response-based pixel data for each image object. The combination of the original image plus the newly created hyperplane is mapped to form a multi-spectral hypercube. The hypercube has pixel dimensions Pn where n is the total number of outputs.
Once a hypercube has been created, a wide range of both spatial and spectral feature characteristics can be measured and extracted – from each individual hyperplane, or using vectors manifested through multiple planes. Machine learning algorithms (MLAs) can now analyze the resulting metrics for each object of interest and compare the results with a “rules-base” created during the development of the application. The objects are then classified by the data analyzer based upon the rules-base.Signature Mapping™ CAD Technology has been developed to provide important clinical tools to improve imaging diagnosis and therapy, these tools include:
- Image Clarification
- Computer Aided Visualization
- Tissue Characterization
- Computer Aided Detection
- Computer Aided Diagnosis
These capabilities build on each other to provide additional clinical information at each level of technical performance.
Additional Information
- Phase I Clinical Studies Completed at USC
- Signature Mapping™ Web PowerPoint Presentation
(best viewed with Internet Explorer 4.0 or higher)
| Webmaster |
Copyright © 2009 Guardian Technologies International. |
