DK1212_C010

DK1212_C010 - 10 Image Fusion This chapter deals with a...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
417 10 Image Fusion This chapter deals with a relatively new direction in image process- ing that proves to be of high importance in many areas, not just in medical applications —obtaining qualitatively new information by combining the image data from several different images (or sequences of images). The term image fusion is used here in a very generic sense of synthesizing the new information based on combining and mutually complementing the information from more than a single image. Although we are dealing largely with two-dimensional images in this book, it should be understood that medical image fusion often concerns three-dimensional image data provided by tomographic modalities. However, the three-dimensional fusion is, disregarding rare exceptions, a conceptually simple generalization of the two- dimensional case. The most common and obviously clinically needed is the image registration in its many variants: intramodality registration and intrapersonal (intrapatient) registration on one side, and intermo- dality and interpersonal registration on the other (the registration of measured image data to atlas images or to physical reality in the interventional radiology also belongs here). Registering images may © 2006 by Taylor & Francis Group, LLC
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
418 Jan enable better comparison of the image information on the same area provided by different modalities, or by a single modality in the course of time, an assessment of anatomy differences inside a group of patients or volunteers, etc. In these cases, registration itself is the aim of processing, and further derivation of conclusions is left to the medical staff. However, even in this simplest case, fusing information from more images leads to a new quality. Registration may also be the first step of a more complex pro- cessing leading to the automatic narrower-sense image fusion serv- ing to derive new images. The simplest and perhaps most frequent example is the subtraction of two images (as in the subtractive angiography) in the hope that the difference image would reveal relevant dissimilarities; the preliminary registration is used to pre- vent enhancing of irrelevant differences due to field-of-view (FOV) shift, differing geometrical distortion, patient movement, etc. The registered images from different modalities may be fused into a single vector-valued image that may serve as a better ground for segmentation or another automatic analysis, or may be presented to the evaluating staff by means of, e.g., false colors or overlay tech- niques. On the other hand, fusion of single-modality multiview images covering contiguous parts of an object and overlapping at margins may enable the forming of images of substantially greater FOV; the fusion usually requires both geometrical and gray-scale (or color) transforms to enable reasonably seamless joining.
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 78

DK1212_C010 - 10 Image Fusion This chapter deals with a...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online