Medical image processing

Abstract

A method of image processing, including: (a) calculating at least one pixel color feature (PCF) value for each pixel in a color medical image to generate a set of PCF data; and (b) filtering the PCF data with at least one spatial adaptive bandpass filter (ABPF) to sort the pixels into physiologically significant regions; wherein the at least one PCF value for at least one pixel depends on at least 2 color components of the medical image.

What is claimed is: 
     
         1 . A method of image processing, the method comprising:
 (a) receiving values for each pixel of an image as digital data;   (b) applying one or more pairs of adaptive smoothing filters to each value, wherein each pair of adaptive smoothing filters has fine and coarse smoothing parameters, and differencing between the pairs of adaptive smoothing filters, thereby producing respective spatial adaptive bandpass filters (ABPFs) for each pixel, whereby said different smoothing parameters define degrees of local contrast with respect to adjacent regions in the image   (c) applying a threshold to each said ABPF thereby detecting image regions of interest, wherein said threshold is one of:
 a global constant threshold determined from global image statistics throughout the image; and 
 a locally adaptive threshold determined from local image statistics throughout the image; 
   wherein said threshold results in a constant false alarm rate (CFAR), thereby reducing the incidence of false positives.   
     
     
         2 . A method according to  claim 1  wherein asymmetry, border irregularity, color variability, and diameter (ABCD) skin lesion criteria are employed to characterize detected lesions. 
     
     
         3 . A method according to  claim 1  wherein skin moles and lesions are characterized in terms of their pigmentation as at least one of black, brown, pink and red. 
     
     
         4 . A method according to  claim 1  wherein low contrast, spread inflammations are enhanced and detected. 
     
     
         5 . A method according to  claim 1  wherein each pixel value includes at least one of intensity, color, infrared, spectral contrast, density of blood vessels, density of lesions, absorbance, reflectance, diffuse reflectance. 
     
     
         6 . A method according to  claim 1  wherein detection maps of a same subject taken at different times are acquired and matched for progressive analysis. 
     
     
         7 . A method according to  claim 1  wherein obstructing hair lines are removed from the image while retaining moles, lesions and blemishes, thereby enabling their detection. 
     
     
         8 . A method according to  claim 1  whereby said ABPFs are applied in several resolution scales. 
     
     
         9 . A method according to  claim 8  whereby a detection map is generated by combining said resolution scales. 
     
     
         10 . A method according to  claim 1  whereby said ABPF processing is applied in generating a body contour. 
     
     
         11 . A method according to  claim 1  whereby said ABPF processing and detection is applied to various skin regions of a body. 
     
     
         12 . A method according to  claim 1  whereby said method is used to align body parts for subsequent analysis of skin abnormalities over time. 
     
     
         13 . A method according to  claim 1  whereby said ABPFs are applied to images acquired by more than one camera, thereby providing distance via triangulation, improved detection and reduced imaging interference. 
     
     
         14 . A method according to  claim 1  whereby said ABPFS are applied in a wide field of view and lower resolution in computing a body or skin contour. 
     
     
         15 . Apparatus for enhancement of a multi-spectral image, said apparatus comprising:
 (a) an input port adapted to receive a multi-spectral image as digital data;   (b) filtration circuitry adapted to apply at least one multi-spectral processing filter to each of a plurality of pixels of the multi-spectral image to evaluate a multi-spectral color feature of each pixel;   (c) sorting circuitry adapted to define at least one multi-spectral image region by applying a spatial adaptive bandpass filter to the enhanced color features of the pixels.   
     
     
         16 . An apparatus according to  claim 15 , wherein said multi-spectral image is a medical or an aesthetic image. 
     
     
         17 . An apparatus according to  claim 15 , wherein said multi-spectral image is one of a medical image, an aesthetic image, a dermal (external) tissue image, and a mucosal (internal) tissue image.