Multiscaled combination of MR and SPECT images in neuroimaging: A simplex method based variable-weight fusion

The relationship of early life adversity and physiological synchrony within the therapeutic triad in horse-assisted therapy

In any therapeutic setting, the outcome depends in part on the therapeutic alliance, characterized by mutual understanding, empathy and trust among the participants. This also manifests through physiological synchronization (PS) processes involving breathing, heart and brain. This study examined the dynamics of heart rate variability (HRV) synchronization patterns during horse-assisted therapy. We explored the correlations between the therapist's horse preference, levels of early life adversity (ELA), and PS relationships within and across dyads of participants, therapists, and therapy horses. Our sample of 42 female participants engaged in standardized horse-assisted therapy sessions facilitated by three riding therapists and four therapy horses. PS was operationalized through cross-wavelet power analyses across the different dyads. The results showed, that stronger HRV synchronization between the therapist and horse was associated with stronger HRV synchronization between the horse and participant, as well as stronger HRV synchronization between the therapist and participant. We found a correlation between ELA and HRV synchronization between participants and therapists, with individuals experiencing higher levels of ELA showing lower synchronization. However, this effect of ELA was not observed for HRV synchronization between participants and horses. Furthermore, we found a negative correlation between the riding therapist's preference for a particular therapy horse and the HRV synchronization between the therapist and that horse. These findings contribute to a better understanding of the correlational dynamics in horse-human interactions and may have potential implications for optimizing therapeutic interventions in clinical settings.

Multimodal neuroimage data fusion based on multikernel learning in personalized medicine

Neuroimaging has been widely used as a diagnostic technique for brain diseases. With the development of artificial intelligence, neuroimaging analysis using intelligent algorithms can capture more image feature patterns than artificial experience-based diagnosis. However, using only single neuroimaging techniques, e.g., magnetic resonance imaging, may omit some significant patterns that may have high relevance to the clinical target. Therefore, so far, combining different types of neuroimaging techniques that provide multimodal data for joint diagnosis has received extensive attention and research in the area of personalized medicine. In this study, based on the regularized label relaxation linear regression model, we propose a multikernel version for multimodal data fusion. The proposed method inherits the merits of the regularized label relaxation linear regression model and also has its own superiority. It can explore complementary patterns across different modal data and pay more attention to the modal data that have more significant patterns. In the experimental study, the proposed method is evaluated in the scenario of Alzheimer’s disease diagnosis. The promising performance indicates that the performance of multimodality fusion via multikernel learning is better than that of single modality. Moreover, the decreased square difference between training and testing performance indicates that overfitting is reduced and hence the generalization ability is improved.

Gaussian Filter for Brain SPECT Imaging

Background. The presence of a noise component on 3D images of single-photon emission computed tomo­graphy (SPECT) of a brain significantly distorts the probability distribution function (PD) of the radioactive count rate in the images. The presence of noise and further filtering of the data, based on a subjective assessment of image quality, have a significant impact on the calculation of volumetric cerebral blood flow and the values of the uptake asymmetry of the radiopharmaceutical in a brain. Objective. We are aimed to develop a method for optimal SPECT filtering of brain images with lipophilic radiopharmaceuticals, based on a Gaussian filter (GF), for subsequent image segmentation by the threshold method.  Methods. SPECT images of the water phantom and the brain of patients with 99mTc-HMPAO were used. We have developed a technique for artificial addition of speckle noise to conditionally flawless data in order to determine the optimal parameters for smoothing SPECT, based on a GF. The quantitative criterion for optimal smoothing was the standard deviation between the PD of radioactive count rate of the smoothed image and conditionally ideal one. Results. It was shown that the maximum radioactive count rate of the SPECT image has an extremum by changing the standard deviation of the GF in the range of 0.3–0.4 pixels. The greater the noise component in the SPECT image, the more quasi-linearly the corresponding rate changes. This dependence allows determining the optimal smoothing parameters. The application of the developed smoothing technique allows restoring the probability distribution function of the radioactive count rate (distribution histogram) with an accuracy up to 5–10%. This provides the possibility to standardize SPECT images of brain. Conclusions. The research results of work solve a specific applied problem: restoration of the histogram of a radiopharmaceuticals distribution in a brain for correct quantitative assessment of regional cerebral blood flow. In contrast to the well-known publications on the filtration of SPECT data, the work takes into account that the initial tomographic data are 3D, rather than 2D slices, and contain not only uniform random Gaussian noise, but also a pronounced speckle component.

MRI and PET image fusion using the nonparametric density model and the theory of variable-weight

Medical image fusion is important in the field of clinical diagnosis because it can improve the availability of information contained in images. Magnetic Resonance Imaging (MRI) provides excellent anatomical details as well as functional information on regional changes in physiology, hemodynamics, and tissue composition. In contrast, although the spatial resolution of Positron Emission Tomography (PET) provides is lower than that an MRI, PET is capable of depicting the tissue's molecular and pathological activities that are not available from MRI. Fusion of MRI and PET may allow us to combine the advantages of both imaging modalities and achieve more precise localization and characterization of abnormalities. Previous image fusion algorithms, based on the estimation theory, assume that all distortions follow Gaussian distribution and are therefore susceptible to the model mismatch problem. To overcome this mismatch problem, we propose a new image fusion method with multi-resolution and nonparametric density models (MRNDM). The RGB space registered from the source multi-modal medical images is first transformed into a generalized intensity-hue-saturation space (GIHS), and then is decomposed into the low- and high-frequency components using the non-subsampled contourlet transform (NSCT). Two different fusion rules, which are based on the nonparametric density model and the theory of variable-weight, are developed and used to fuse low- and high-frequency coefficients. The fused images are constructed by performing the inverse of the NSCT operation with all composite coefficients. Our experimental results demonstrate that the quality of images fused from PET and MRI brain images using our proposed method MRNDM is higher than that of those fused using six previous fusion methods.

Imaging and machine learning techniques for diagnosis of Alzheimer's disease

Alzheimer's disease (AD) is a common health problem in elderly people. There has been considerable research toward the diagnosis and early detection of this disease in the past decade. The sensitivity of biomarkers and the accuracy of the detection techniques have been defined to be the key to an accurate diagnosis. This paper presents a state-of-the-art review of the research performed on the diagnosis of AD based on imaging and machine learning techniques. Different segmentation and machine learning techniques used for the diagnosis of AD are reviewed including thresholding, supervised and unsupervised learning, probabilistic techniques, Atlas-based approaches, and fusion of different image modalities. More recent and powerful classification techniques such as the enhanced probabilistic neural network of Ahmadlou and Adeli should be investigated with the goal of improving the diagnosis accuracy. A combination of different image modalities can help improve the diagnosis accuracy rate. Research is needed on the combination of modalities to discover multi-modal biomarkers.

The Nonsubsampled Contourlet Transform Based Statistical Medical Image Fusion Using Generalized Gaussian Density

We propose a novel medical image fusion scheme based on the statistical dependencies between coefficients in the nonsubsampled contourlet transform (NSCT) domain, in which the probability density function of the NSCT coefficients is concisely fitted using generalized Gaussian density (GGD), as well as the similarity measurement of two subbands is accurately computed by Jensen-Shannon divergence of two GGDs. To preserve more useful information from source images, the new fusion rules are developed to combine the subbands with the varied frequencies. That is, the low frequency subbands are fused by utilizing two activity measures based on the regional standard deviation and Shannon entropy and the high frequency subbands are merged together via weight maps which are determined by the saliency values of pixels. The experimental results demonstrate that the proposed method significantly outperforms the conventional NSCT based medical image fusion approaches in both visual perception and evaluation indices.

Adaptive registration of magnetic resonance images based on a viscous fluid model

This paper develops a new viscous fluid registration algorithm that makes use of a closed incompressible viscous fluid model associated with mutual information. In our approach, we treat the image pixels as the fluid elements of a viscous fluid governed by the nonlinear Navier-Stokes partial differential equation (PDE) that varies in both temporal and spatial domains. We replace the pressure term with an image-based body force to guide the transformation that is weighted by the mutual information between the template and reference images. A computationally efficient algorithm with staggered grids is introduced to obtain stable solutions of this modified PDE for transformation. The registration process of updating the body force, the velocity and deformation fields is repeated until the mutual information reaches a prescribed threshold. We have evaluated this new algorithm in a number of synthetic and medical images. As consistent with the theory of the viscous fluid model, we found that our method faithfully transformed the template images into the reference images based on the intensity flow. Experimental results indicated that the proposed scheme achieved stable registrations and accurate transformations, which is of potential in large-scale medical image deformation applications.