Interactive Feature Space Explorer

By their local dependence on different biological, chemical and structural tissue properties, biomedical imaging signal generating mechanisms provide the detection of tissue transformations for diagnostic and/or prognostic purposes. Whereas multimodal imaging improves inaccuracies originating from signal overlaps between malignant and benign conditions, it creates a data deluge with inter– and intra–observer bias.

Here a new computer aided approach based on the feature space, which describes tissue properties in a single instance, is proposed to establish an evidence based, concise, direct and objective methodology. The aim is the discovery of appropriate imaging modalities and biomarker value of their combinations by introducing the experience of physicians in the discovery process. The platform allows the user to interactively compare the features and the corresponding anatomic loci in order to identify malignant FS regions. Results on a potential aggressiveness imaging biomarker in prostate cancer and preliminary findings on renal cell carcinoma are presented.

This invention can be used in biopsies, image guided therapies, robotic surgery, and diagnosis of cancer or other diseases of different organs such as prostate, kidney and brain. Additionally, this invention discloses a method of reduction in visualization of large amounts of data by looking at the feature space to evaluate the situation rather than having to go through a large number of images.

Patent Information:
For Information, Contact:
Li Chen
Licensing Associate
Virginia Tech Intellectual Properties, Inc.
(540) 443-9217
Alpay Ozcan