Deformable group-wise registration using a physiological model: Application to diffusion-weighted MRI

Abstract

Intensity variations can often be described by a physiological or temporal model applied on a voxel-wise basis across a group of images. However the voxel correspondence might be unknown, imposing the need for a group-wise deformable registration coupled with the computation of the model parameters. In this paper we propose a group-wise registration method of medical images that incorporates the temporal dimension (reflecting the change of signal amplitude) of the acquisition process. Consistency on the spatiotemporal physiological model, as well as deformation smoothness, is imposed in order to produce anatomically meaningful representations of the 3D images. The performance of the proposed method is compared to two different group-wise registration approaches; one that penalizes the absolute differences in the intensities and one that penalizes the intensity range among the images on corresponding regions. We chose as an application paradigm the registration of diffusion-weighted magnetic resonance (DW-MR) images for the evaluation of patients with lymphomas. A dataset consisting of 25 patients, each scanned with 3 "b values", was used to evaluate the method's accuracy. The proposed registration method outperfomed the other two registration approaches, making it a very promising method for highlighting the importance of DWI as an imaging biomarker.

On motion in dynamic magnetic resonance imaging: Applications in cardiac function and abdominal diffusion

“…(3.2) ( S j = S 0 exp(−b j ADC i )) for a given b-value j out of a total of N b . The proposed metric extends the one presented in Kornapoulos et al (2016) by adding proper weighting parameters related to the underlying noise distribution of the magnitude images. Those weights, initially unitary, are redefined through iterations along with the predicted DW signals from ADC estimates as:…”

“…GW approaches have been addressed in Veeranghavan et al (2015) using prior structure segmentations, although the intensity changes due to the diffusion process are not considered. GW approaches that use a Markov Random fields (MRF) have also been described in Kornapoulos et al (2016). Recently, spatiallyconstrained approaches have been proposed for liver DWI nonrigid registration (see Kurugol et al (2017b); Taimouri et al (2015)) grounded on MRF.…”

“…In Kornapoulos et al (2016); Kurugol et al (2017a), it is hypothesized that registration metrics that explicitly take into account the diffusion model and, consequently, the parameters to be estimated will not only make the ulterior ADC analysis more robust but will also alleviate the effects of the additional confounding factors. Different approaches have been described in the literature focused on the improvement of the estimation accuracy of diffusion parameters, for example, Kalman filters (Poupon et al, 2008) or the Linear Minimum Mean Squared Error (LMMSE) framework .…”

“…All datasets, both real and synthetic, have been processed with different GW multimodal metrics (Sanz-Estébanez et al, 2017) as well as different registration methodologies. As for the former, apart from the one here proposed (bWei, from b-Weighted), we have tried EDI (Learned-Miller, 2006), MIND (Heinrich et al, 2011) and the one proposed in Kornapoulos et al (2016) based on non-weighted ADC residuals (ADC). Many more metrics have been proposed to tackle this problem (see Glocker et al (2011)); however, from our point of view, the ones here proposed compose a sufficiently representative sample.…”

“…GW approaches have been addressed in Veeranghavan et al (2015) using prior structure segmentations, although the intrinsic intensity changes due to the diffusion process are not considered. Hence, we hypothesize that a registration metric that explicitly takes into account the parameters to be estimated, as proposed in Kornapoulos et al (2016); Kurugol et al (2017a), will not only make the ulterior ADC analysis more robust but will also alleviate the effects of the different confounding factors.…”