Technique for manipulating medical images
AssigneeThe Board of Trustees of the Leland Stanford Junior University
InventorsDaniel Steines, Philipp Lang
Life of this patent
Estimating the term…
Abstract
The invention and the embodiments described in this invention provide new techniques for manipulating digital images and is particularly useful for extracting tissues (i.e., assigning tissue boundary locations) from medical images. These techniques can be applied to diagnosing arthritis and for monitoring disease progression or response to therapeutic intervention. The invention provides for means to extract the articular cartilage from medical images for analysis purposes.
Claims15
A method of assessing a digital image to determine a boundary between first and second image components, which comprises:
- (a) selecting a first boundary element between two pixels being evaluated for assignment to the first or second image component;
- (b) determining a first value of a first pixel on one side of the first boundary element;
- (c) determining a second value of a second pixel on the other side of the first boundary element;
- (d) comparing the first and second value to determine magnitude of a gradient across the first boundary element and assigning a gradient boundary value to the first boundary element;
- (e) determining relative locations of (1) the first boundary element and (2) a reference element selected from the group consisting of (2a) a reference boundary element previously assigned as separating a pixel assigned to the first image component from a pixel assigned to the second image component and (2b) a reference second-boundary element previously assigned as separating a pixel assigned to one of the first and second image components from a third image component, the second-boundary element having a known, statistically significant distance or angle relationship relative to an actual boundary between real components represented by the first and second image components, and assigning a location boundary value to the first boundary element based on distance or angle of the relative locations from each other;
- (f) mathematically combining the gradient boundary value and the location boundary value to provide a first potential boundary value for the first boundary element;
- repeating steps (a)-(f) for a second boundary element to obtain a second potential boundary value;
- and
- (g) assigning the first boundary element or the second boundary element as dividing the first image component from the second image component, depending on relative potential boundary values of the first and second boundary elements, wherein said method is used to locate a bone/cartilage interface or an external cartilage surface.
Dependencies are inferred from the claim text.