Sti Medical Systems, Inc.

Sankar Venkataraman

Ulf Peter Gustafsson

Wenjing Li

John Taylor Hargrove

A method of contrast enhancement for improved visualization of diagnostically important tissue structures, such as blood vessels. A texture analysis algorithm is applied to identify regions with a high likelihood of disease. Mathematical morphology operations are applied to identify areas of high and low brightness (intensity). The low intensity areas are then subtracted, and controllably variable amounts of the high intensity areas are added, controlled by a selectable tuning parameter, to produce an image with controllably variable visualization enhancement.

Methods of contrast enhancement for images having blood vessel structures

The present invention discloses a method for the detection of atypical vessels in digital cervical imagery. A pre-processing stage is applied to enhance the contrast of blood vessel features compared to the surrounding tissue. Next, a segmentation stage is applied to identify regions of interest for atypical vessels using texture and gradient information. Finally, a post-processing stage (1) identifies other clinically relevant features in the cervical imager, and removes these features from the region of interest; and (2) uses color, size, and shape information to further refine the region of interest to eliminate false positives and determine a final region of interest. This automated method of atypical vessel detection is especially useful for diagnostic purposes such as cervical cancer detection.

Growing community of inventors

Honolulu, HI, United States of America

Sankar Venkataraman