ALIGNMENT OF MULTIMODAL NOISY IMAGES - Isis Project No 1405

Isis Innovation, the technology transfer arm of the University of Oxford, releases a new method for automatically aligning multimodal noisy images.

Marketing Opportunity

The alignment of two or more images is a fundamental image analysis problem with a diverse set of applications.  These include Image registration, for example medical image registration, in which the images may be of the same or different types (e.g. MRI, CT, PET in the case of medicine; infrared, visual, synthetic aperture radar in the case of aerial image analysis; etc) or may be of the same or different subjects (patients) at different times. Alternatively, the challenge may include matching two images (for example aerial images) of nominally the same scene; but which differ either because they are taken with different sensors or frequency bands (e.g. Infrared or ultraviolet) or under different imaging conditions. In the case of image motion, the computation of the optic flow or the estimation of three-dimensional structure from motion, require that images be aligned. Finally, in the case of multi-view image analysis such as stereovision, the simultaneously taken images from two or more cameras are combined to form an estimate of the three dimensional locations of points in the scene.

Although primarily aimed at medical imaging (2, 3 and 4 dimensional) applications this technology could be used in any of the imaging fields described.

The Oxford Invention

Image alignment is a difficult problem for several reasons; the signal to noise ratio of many images, particularly medical images, is low; the two (or more) images to be aligned often have different spatial resolutions (e.g. CT and PET); and the objects to be aligned in the two images may require “warping” from one image to the other. The Oxford invention uses a new similarity measure as well as a process for coping with noise content that delivers a technique that supersedes current image alignment methodologies.

Other Medical Imaging Projects

1260 - Breast Cancer Detection
1283 - Image Velocity Estimation in Echocardiographs
1487 - Lung Nodules Detection in CT Images
2158 - Automated Retinal Image Differencing (ARID)
2468 - Targeted Image Contrast Agent
2586 - Optimal Whole-Heart Ultrasound Imaging

Request Further Information on Project 1405 - Alignment of Multimodal Noisy Image