Wound imaging and analysis

Inventors

DACOSTA, Ralph • MEANEY, Todd E. • Daynes, Todd • VERMEY, Garrett R. • Teene, Liis • DUNHAM, Danielle C. • ABHARI, Kamyar • McFADDEN, Steven

Assignees

Moleculight Inc

Abstract

Given a specific imaging device and systems further described herein, wound characteristics of a wound fluoresce with a unique spectral signature when subjected to excitation light with a known wavelength or range of wavelengths. Images captured therefrom are subject to analyses of pixels thereof, with a plurality of training images having known wound sizes and characteristics marked-up thereon being used to generate training data, which is subsequently used to identify wound characteristics from test images in real time. Wound sizes, boundaries, bacterial presence, and other characteristics may be quantified and graphically represented as an overlay on the original wound image along with documentation related to the wound.

Core Innovation

The invention provides a computer-implemented wound analysis that uses fluorescence-based wound imaging conceptually represented as images in which a wound, and optionally related tissue/tumor content, exhibits a detectable spectral/color behavior under illumination. An image of a wound comprising a plurality of pixels is processed by applying a chroma mask on a pixel-by-pixel basis, where the chroma mask is based on a histogram of pixel values, to determine at least one area of interest comprising one or more wound characteristics.

After determining the at least one area of interest using the chroma mask and histogram of pixel values, one or more contours are detected around the at least one area of interest or defining one or more boundaries around the at least one area of interest. The method then generates an output image comprising the one or more contours overlaid on the image, producing a composite image identifying the at least one area of interest and presented to a user, including in real time.

In further embodiments, the invention expands the training and identification of spectral signatures by converting RGB images into a hue-saturation-value color space and determining histograms of HSV values for known areas of interest. Composite histograms are generated from the histograms of HSV values for multiple known areas of interest so that the composite histogram is used to identify unknown areas of interest in real time on a wound imaging device based on one or more unique spectral signatures. In additional workflows, the invention detects at least one marker in an image and registers the image based on detected markers.

Claims Coverage

The partial set includes four independent claims. Across these claims, the inventive features cover chroma-mask-and-histogram wound area identification, contour detection with overlaid composite output, histogram training using HSV unique spectral signatures to identify unknown areas in real time, and marker detection with image registration using thresholding and connected-component grouping.

Overall, the independent claims focus on determining wound or wound/tissue areas of interest using a chroma mask derived from pixel-value histograms, identifying boundaries via contour detection and generating composite overlay outputs, training and applying HSV histogram-based unique spectral signatures to identify unknown areas of interest in real time, and using threshold-based marker detection with connected-component analysis to register images.

Stated Advantages

Documented Applications