NUMERICS: An online image registration and image comparison platform

The objective of this work is to present a newly developed image registration and image comparison web platform called NUMERICS. The platform acts as an interface between the user and a set of dedicated original algorithms for image comparison of noisy medical images and image registration. The main purpose of the platform is to provide a user-friendly environment for the application of various image processing algorithms. The platform consists of two main components a website and a server-side application. The web-site is freely accessible and offers an intuitive user interface for uploading pairs of 2D images and for adjusting the parameters of the image processing algorithms. The actual computation is performed by the server-side application. The obtained results from the computation are displayed back to the user on a separate web page. The developed platform offers a wide set of image processing operations for 2D images. The user can choose between different types of scaling and transformation of the pixel intensities in the original images. The implemented image registration algorithms include basic methods such as rigid registration, non-rigid registration with affine transformations and thin-plate spline deformable registration. A more advanced approach for image registration, based on Mobius transformations in the complex plane, is also available. The platform also allows online image comparison of image pairs via dedicated image comparison algorithms based on the Grubbs statistical test for outliers. All of the algorithms are specially designed for application to noisy medical images. The platform can find application in PET imaging and diagnosis, SPECT heart imaging and diagnosis and can also serve as a tool for performing quantitative evaluation of the performance of different image processing techniques.