Image Processing Tools for Biomedical Infrared Imaging

Image Processing Tools for Biomedical Infrared Imaging

Gerald Schaefer (Loughborough University, UK) and Arcangelo Merla (University G. D’Annunzio – Chieti-Pescara, Italy)
DOI: 10.4018/978-1-60566-768-3.ch012
OnDemand PDF Download:
List Price: $37.50


Medical infrared imaging captures the temperature distribution of the human skin and is employed in various medical applications. Unfortunately, many of the conventional and commercial suites for image processing provide only very basic tools for the processing of medical thermal images which represent a challenging combination of both functional and morpho-structural imaging. In this chapter, several more advanced approaches are discussed which in turn provide tremendous help to the clinician. As an example, it is often useful to cross-reference thermograms with visual images of the patient, either to see which part of the anatomy is affected by a certain disease or to judge the efficacy of the treatment. It is shown that image registration techniques can be effectively used to generate an overlay of visual and thermal images to provide a useful diagnostic visualisation. Image registration can also be performed based on two thermograms and a warping-based method for this is presented. Segmenting the background from the foreground (i.e., the patient) is a crucial task and it is highlighted how this can be accomplished. Finally, it is shown how descriptors, extracted from medical infrared images, can be usefully employed to search through a large database of cases as well as to aid in diagnosis.
Chapter Preview

Thermal-Visual Image Overlays

Often visual and infrared images of the patient are taken to relate inflamed skin areas to the human anatomy which is useful for medial diagnosis as well as for assessing the efficacy of any treatment. Currently this process requires great expertise and is subject to the individual clinician’s ability to mentally map the two distinctly different images. Therefore, an overlay of the two image types resulting in a composite image which makes it possible to cross-reference regions with unusual temperature distributions to the human anatomy will provide a useful tool for improved medical diagnosis.

Such an overlay can be achieved through application of an image registration technique (Tait et al., 2006). Registration is a method used to geometrically align or overlay two images taken from different sensors, viewpoints or instances in time (Zitova & Flusser, 2003). A reference (fixed) and a sensed (moving) image are aligned through a combination of scaling, translation and rotation, i.e. through an affine transform which is also the type of transform that we employ in our approach. Registration techniques can typically be classified as either intensity or landmark-based. Both techniques have advantages and disadvantages in their own unique approach. The main difficulty of landmark-based algorithms is the need to identify a set of corresponding control points in both images based upon which the best matching transform is sought. Landmarks can be found either manually or automatically. While manual selection of control points can be fairly time consuming and requires user interaction, automatic identification of landmarks constitutes a challenging problem and often requires a priori knowledge of the image features involved.

Complete Chapter List

Search this Book: