Abstract-

Abstract. In order to improve the robustness of rigid registration algorithms in various medical imaging problems, we propose in this article a general framework built on block matching strategies. This framework combines two stages in a multi-scale hierarchy. The first stage consists in finding for each block (or subregion) of the first image, the most similar subregion in the other image, using a similarity criterion which depends on the nature of the images. The second stage consists in finding the global rigid transformation which best explains most of these local correspondances. This is done with a robust procedure which allows up to 50% of false matches. We show that this approach, besides its simplicity, provides a robust and efficient way to rigidly register images in various situations. This includes for instance the alignment of 2D histological sections for the 3D reconstructions of trimmed organs and tissues, the automatic computation of the mid-sagittal plane in multimodal 3D images of the brain, and the multimodal registration of 3D CT and MR images of the brain. A quantitative evaluation of the results is provided for this last example, as well as a comparison with the classical approaches involving the minimization of a global measure of similarity based on Mutual Information or the Correlation Ratio. This shows a significant improvement of the robustness, for a comparable final accuracy. Although slightly more expensive in terms of computational requirements, the proposed approach can easily be implemented on a parallel architecture, which opens potentialities for real time applications using a large number of processors.

Diffusion-Weighted MRI (DW-MRI) is subject to random noise yielding measures that are different from their real values, and thus biasing the subsequently estimated tensors. The Non-Local Means (NLMeans) filter has recently been proposed to denoise MRI with high signal-to-noise ratio (SNR). This filter has been shown to allow the best restoration of image intensities for the estimation of diffusion tensors (DT) compared to state-of-the-art methods. However, for DW-MR images with high b-values (and thus low SNR), the noise, which is strictly Rician-distributed, can no longer be approximated as additive white Gaussian, as implicitly assumed in the classical formulation of the NLMeans. High b-values are typically used in high angular resolution diffusion imaging (HARDI) or q-space imaging (QSI), for which an optimal restoration is critical. In this paper, we propose to adapt the NLMeans filter to Rician noise corrupted data. Validation is performed on synthetic data and on real data for both conventional MR images and DT images. Our adaptation outperforms the original NLMeans filter in terms of peak-signal-to-noise ratio (PSNR) for DW-MRI.