miRID: Multi-Modal Image Registration Using Modality-Independent and Rotation-Invariant Descriptor

Abstract

Axiomatically, symmetry is a fundamental property of mathematical functions defining similarity measures, where similarity measures are important tools in many areas of computer science, including machine learning and image processing. In this paper, we investigate a new technique to measure the similarity between two images, a fixed image and a moving image, in multi-modal image registration (MIR). MIR in medical image processing is essential and useful in diagnosis and therapy guidance, but still a very challenging task due to the lack of robustness against the rotational variance in the image transformation process. Our investigation leads to a novel, local self-similarity descriptor, called the modality-independent and rotation-invariant descriptor (miRID). By relying on the mean of the intensity values, an miRID is simply computable and can effectively handle the complicated intensity relationship between multi-modal images. Moreover, it can also overcome the problem of rotational variance by sorting the numerical values, each of which is the absolute difference between each pixel’s intensity and the mean of all pixel intensities within a patch of the image. The experimental result shows that our method outperforms others in both multi-modal rigid and non-rigid image registrations.

Multi-modal Brain MR Image Registration using A Novel Local Binary Descriptor based on Statistical Approach

“…Its advantages are simplicity of computation and effectiveness in dealing with different modalities. This section will investigate the performance, in terms of registration accuracy, of the proposed algorithm in comparison with other approaches: MI [38], SSC [18], miLBP [25], RWMI [30], and miRID [32]. This experiments were configured the image patch size as follows: for the MI and RWMI methods, the image patch with the size of 7 x 7 pixels and 64 bins was used.…”

“…However, in this experimentation on the registrations of T1-T2, T1-PD, and T2-PD using various approaches, it has been found that the proposed method, SLBD, has the least errors. In comparison, the miLBP [25], RSSD [31], and miRID [32] have the high speeds of MIR and also have the robustness of modality-independent. The miRID [32] could perform the best performance in MIR both rigid and non-rigid registration.…”

“…In comparison, the miLBP [25], RSSD [31], and miRID [32] have the high speeds of MIR and also have the robustness of modality-independent. The miRID [32] could perform the best performance in MIR both rigid and non-rigid registration. However, those descriptors may lose the structural information due to their calculation.…”

“…, with ( 4) miRID [32] has the potential to increase the stability of medical image registration to rotation variations. Instead of utilizing the intensity of the center pixel, miRID uses the mean of all intensity values inside the patch to achieve this.…”

“…The other work of authors also enhanced the miLBP, which adopts the mean and standard deviation of the image patch for adaptive threshold and uses the sorting operation for sorting the pixels within image patch. Therefore, it could provide terms of robustness in modalityindependent and rotation-invariant descriptor (miRID) [32]. Yang et al [33] have proposed MIR based on image segmentation and symmetric self-similarity.…”