Color-image processing: An introduction with some medical application-examples

摘要

Color is that attribute of light-energy which is related to the wavelength. It is well known that color carries a very important part of information regarding objects of interest in an image. This talk introduces the audience to the representation of color (color-models) and color-based segmentation of images, with several illustrative examples from the speaker's research-experience in this area over the past few years. Application-examples considered in this talk include (i) wound-image processing, (ii) TB-screening, and (iii) Segmentation of Fascia. The first step in the treatment of wounds is that of visual assessment of the wound-site. Of the various parameters, color is an important indicator of wound-composition, due to one-to-one correspondence between the tissue-type/pigmentation and color. Since visual assessment is tedious and subjective, digital color-image processing is helpful in rendering the process automatic and also allow documentation and monitoring. Color-composition within a wound-site can be assessed by building a data-base of wound-color-types in the color-space, and using the same for classification upon the presentation of a new wound. A total of eight types of tissue/pigmentation have been shown to exist. Tuberculosis (TB) is a serious illness that can result in mortality if neglected. Pulmonary TB is the most infectious, and is generally detected by a visual examination of highly magnified images of stained sputum smear. Since visual examination is tedious and can result errors, it would be useful to develop an automatic TB screening method. The method of color-segmentation described in the case of wound-image processing works on images of ZN-stained sputum smears, towards segmenting candidate bacterial structures from the background. After color-based segmentation, certain post-processing techniques allow a counting of the number of bacilli in images of stained sputum smear, which provide the degree of infection. Human abdominal-fascia co--nsists of collagen and elastic fibers, and knowing the relative content of these fibers can be useful in plastic surgery. It has been shown that a convenient color-model exists, that can be used for segmenting the two types of fibers from an appropriately stained fascia.