Information Similarity Measure Research Articles

Performance of mutual information similarity measure for registration of multitemporal remote sensing images

Accurate registration of multitemporal remote sensing images is essential for various change detection applications. Mutual information has recently been used as a similarity measure for registration of medical images because of its generality and high accuracy. Its application in remote sensing is relatively new. There are a number of algorithms for the estimation of joint histograms to compute mutual information, but they may suffer from interpolation-induced artifacts under certain conditions. In this paper, we investigate the use of a new joint histogram estimation algorithm called generalized partial volume estimation (GPVE) for computing mutual information to register multitemporal remote sensing images. The experimental results show that higher order GPVE algorithms have the ability to significantly reduce interpolation-induced artifacts. In addition, mutual-information-based image registration performed using the GPVE algorithm produces better registration consistency than the other two popular similarity measures, namely, mean squared difference (MSD) and normalized cross correlation (NCC), used for the registration of multitemporal remote sensing images.

Mutual information based respiration detection

Abstract Percutaneous Transluminal Coronary Angioplasty (PTCA) is currently the preferred method for the treatment of coronary artery disease in vivo. During the intervention, images of the coronary arteries filled with a radiopaque contrast agent are acquired (angiography) in order to localize the lesion. One of these images is then displayed on a monitor. It serves as a roadmap that is afterwards utilised to navigate the interventional instruments to the lesion under constant low-dose X-ray surveillance (fluoroscopy). The fluoroscopic live images are displayed in real-time on a second monitor next to the roadmap monitor. It is desirable to provide for each fluoroscopic image an angiographic roadmap that exhibits a similar position and shape of the coronary vessels. To accomplish this goal, the respiration status in every (pre-recorded) angiographic and (live) fluoroscopic image has to be quantified in order to display angiographic roadmaps matching to the live fluoroscopic image. We present a method that estimates the respiration status of a patient from grey-level X-ray fluoroscopic images based on their similarity to a reference frame applying a mutual information (MI) similarity measure.

Reduction of motion artifacts in three-dimensional CT-DSA using constrained adaptive multilevel free-form registration

In this paper, we describe the application of an intensity-based nonrigid registration algorithm to (3-D) CT-digital subtraction angiography (CT-DSA). On a uniform control point grid (CPG), a B-spline deformation is computed, which maximizes the normalized mutual information (NMI) similarity measure between the 3-D pre- and post-contrast CT images. This transformation is then applied when computing subtraction images. We demonstrate that this algorithm can measure nonrigid patient motion between the image acquisitions and substantially reduces motion artifacts relative to results obtained with rigid-body registration.

On the relationship between the efficiency measures of multicategory information systems

The concepts of efficiency measures of multicategory information systems and similarity measures between two efficiency measures are introduced and developed. The relationship between two efficiency measures is defined by the corresponding set D , and the appropriate marginal functions are introduced. An earlier work of Ben-Bassat's dealing with the ranking of features according to feature selection criteria is generalized and refined. Consequently, a general similarity measure between two efficiency measures is established. It is analyzed in the case of monotonic marginal functions and, especially, when the set D is convex. Some important sufficient conditions for D to be convex are also given. Finally, two other similarity measures are defined and interpreted.