Skin abnormality monitoring systems and methods
A method of monitoring skin abnormalities on a skin portion of a patient including the steps of receiving a first image data from an image capture device and identifying a first skin mask corresponding to the skin portion. The image data includes the skin portion and a first scaling element on the skin portion. The method further includes the steps of identifying one or more first keypoints within the first skin mask and classifying the one or more keypoints. Each first keypoint comprises an abnormality on the skin portion.
1. A method of monitoring skin abnormalities on a skin portion of a patient, comprising the steps of:
receiving a first image data from an image capture device, wherein the image data includes the skin portion and a first scaling element on the skin portion;
identifying a first skin mask corresponding to the skin portion;
identifying one or more first keypoints within the first skin mask, wherein each first keypoint comprises an abnormality on the skin portion;
determining one or more body landmarks on the skin portion;
receiving a second image data from an image capture device, wherein the second image data includes at least a part of the skin portion and a second scaling element on the skin portion, and wherein the first and second image data are sequentially captured;
identifying a second skin mask corresponding to the second skin portion;
identifying one or more second keypoints within the second skin mask, wherein each second keypoint comprises an abnormality on the skin portion;
comparing the one or more first keypoints with the one or more second keypoints of the skin portion, utilizing the one or more body landmarks and the first and second scaling elements to match the skin mask of the first and second image data.
2. The method of claim 1 , further comprising the steps of:
classifying the one or more first keypoints; and
classifying the one or more second keypoints.
3. The method of claim 2 , wherein an abnormality is a pigmented lesion.
4. The method of claim 2 , wherein the first and second keypoints are determined using the Laplacian-of-Gaussian filters.
5. The method of claim 2 , wherein the first and second keypoints are classified into categories selected from the group consisting of: mole, skin, edge, background, hair, fabrics and freckles.
6. The method of claim 2 , wherein two or more of the first keypoints are duplicates, and duplicate keypoints are removed prior to the classifying step.
7. The method of claim 2 , further comprising the step of merging at least two keypoints.
8. The method of claim 7 , wherein the at least two keypoints includes a first detection data and a second detection data, and
wherein the step of merging includes obtaining a first cropped image of the first detection data, a second cropped image of the second detection data, and a third cropped image of the first and second detection data,
wherein the merging step comprises the step of performing a contour analysis of on each of the first, second, and third cropped images, and wherein if the third cropped image contains a single contour, the first and second detection data is merged.
9. The method of claim 2 , further comprising the step of computing a geometrical property for each first and second keypoints.
10. The method of claim 9 , wherein the geometrical property includes one or more of the following properties: height, width, perimeter, area, color distribution, and a dominant color.
11. The method of claim 2 , wherein at least one of the steps of classifying keypoints data, determining body landmarks, merging keypoint data, and matching keypoints is repeated.