Matthew Nitzken

3D shape analysis for early diagnosis of malignant lung nodules

An alternative method of diagnosing malignant lung nodules by their shape, rather than conventional growth rate, is proposed. The 3D surfaces of the detected lung nodules are delineated by spherical harmonic analysis that represents a 3D surface of the lung nodule supported by the unit sphere with a linear combination of special basis functions, called Spherical Harmonics (SHs). The proposed 3D shape analysis is carried out in five steps: (i) 3D lung nodule segmentation with a deformable 3D boundary controlled by a new prior visual appearance model; (ii) 3D Delaunay triangulation to construct a 3D mesh model of the segmented lung nodule surface; (iii) mapping this model to the unit sphere; (iv) computing the SHs for the surface; and (v) determining the number of the SHs to delineate the lung nodule. We describe the lung nodule shape complexity with a new shape index, the estimated number of the SHs, and use it for the K-nearest classification into malignant and benign lung nodules. Preliminary experiments on 327 lung nodules (153 malignant and 174 benign) resulted in a classification accuracy of 93.6%, showing that the proposed method is a promising supplement to current technologies for the early diagnosis of lung cancer.

Local wavelet-based filtering of electromyographic signals to eliminate the electrocardiographic-induced artifacts in patients with spinal cord injury

Surface Electromyography (EMG) is a standard method used in clinical practice and research to assess motor function in order to help with the diagnosis of neuromuscular pathology in human and animal models. EMG recorded from trunk muscles involved in the activity of breathing can be used as a direct measure of respiratory motor function in patients with spinal cord injury (SCI) or other disorders associated with motor control deficits. However, EMG potentials recorded from these muscles are often contaminated with heart-induced electrocardiographic (ECG) signals. Elimination of these artifacts plays a critical role in the precise measure of the respiratory muscle electrical activity. This study was undertaken to find an optimal approach to eliminate the ECG artifacts from EMG recordings. Conventional global filtering can be used to decrease the ECG-induced artifact. However, this method can alter the EMG signal and changes physiologically relevant information. We hypothesize that, unlike global filtering, localized removal of ECG artifacts will not change the original EMG signals. We develop an approach to remove the ECG artifacts without altering the amplitude and frequency components of the EMG signal by using an externally recorded ECG signal as a mask to locate areas of the ECG spikes within EMG data. These segments containing ECG spikes were decomposed into 128 sub-wavelets by a custom-scaled Morlet Wavelet Transform. The ECG-related sub-wavelets at the ECG spike location were removed and a de-noised EMG signal was reconstructed. Validity of the proposed method was proven using mathematical simulated synthetic signals and EMG obtained from SCI patients. We compare the Root-mean Square Error and the Relative Change in Variance between this method, global, notch and adaptive filters. The results show that the localized wavelet-based filtering has the benefit of not introducing error in the native EMG signal and accurately removing ECG artifacts from EMG signals.

Spherical harmonic analysis of cortical complexity in autism and dyslexia

Alterations in gyral form and complexity have been consistently noted in both autism and dyslexia. In this present study, we apply spherical harmonics, an established technique which we have exapted to estimate surface complexity of the brain, in order to identify abnormalities in gyrification between autistics, dyslexics, and controls. On the order of absolute surface complexity, autism exhibits the most extreme phenotype, controls occupy the intermediate ranges, and dyslexics exhibit lesser surface complexity. Here, we synthesize our findings which demarcate these three groups and review how factors controlling neocortical proliferation and neuronal migration may lead to these distinctive phenotypes.

3D shape analysis of the brain cortex with application to autism

To discriminate more accurately between autistic and normal brains, we detect the brain cortex variability using a spherical harmonic analysis that represents a 3D surface supported by the unit sphere with a linear combination of special basis functions, called spherical harmonics (SHs). The proposed 3D shape analysis is carried out in five steps: (i) 3D brain cortex segmentation, with a deformable 3D boundary, controlled by two probabilistic visual appearance models (the learned prior and the estimated current appearance one); (ii) 3D Delaunay triangulation to construct a 3D mesh model of the brain cortex surface; (iii) mapping this model to the unit sphere; (iv) computing the SHs for the surface; and (v) determining the number of the SHs to delineate the brain cortex. We describe the brain shape complexity with a new shape index, the estimated number of the SHs, and use it for K-nearest classification of normal and autistic brains. Initial experiments suggest that our shape index is a promising supplement to the current autism diagnostic techniques.

Automatic analysis of left ventricle wall thickness using short-axis cine CMR images

A new automatic framework for analyzing wall thickness and thickening function on short-axis cine cardiac magnetic resonance (CMR) images is proposed. Inner and outer wall borders (contours) are segmented in a CMR image with a level set deformable model. Its evolution is controlled by a stochastic speed function that accounts for an “object-background” Markov-Gibbs shape and appearance model. Found by solving a Laplace equation, point-to-point correspondences between the inner and outer borders provide initial estimates of the local wall thickness and thickening function index. Effects of segmentation errors are reduced and a 3-D continuity analysis of the left ventricle (LV) wall thickening values is performed by using the maximum a posteriori (MAP) estimates for a pairwise energy function of a generalized Gauss-Markov random field (GGMRF) probabilistic model. Experiments with in vivo CMR data confirm the robustness and accuracy of the proposed framework.

Infant Brain Extraction in T1-Weighted MR Images Using BET and Refinement Using LCDG and MGRF Models

In this paper, we propose a novel framework for the automated extraction of the brain from T1-weighted MR images. The proposed approach is primarily based on the integration of a stochastic model [a two-level Markov-Gibbs random field (MGRF)] that serves to learn the visual appearance of the brain texture, and a geometric model (the brain isosurfaces) that preserves the brain geometry during the extraction process. The proposed framework consists of three main steps: 1) Following bias correction of the brain, a new three-dimensional (3-D) MGRF having a 26-pairwise interaction model is applied to enhance the homogeneity of MR images and preserve the 3-D edges between different brain tissues. 2) The nonbrain tissue found in the MR images is initially removed using the brain extraction tool (BET), and then the brain is parceled to nested isosurfaces using a fast marching level set method. 3) Finally, a classification step is applied in order to accurately remove the remaining parts of the skull without distorting the brain geometry. The classification of each voxel found on the isosurfaces is made based on the first- and second-order visual appearance features. The first-order visual appearance is estimated using a linear combination of discrete Gaussians (LCDG) to model the intensity distribution of the brain signals. The second-order visual appearance is constructed using an MGRF model with analytically estimated parameters. The fusion of the LCDG and MGRF, along with their analytical estimation, allows the approach to be fast and accurate for use in clinical applications. The proposed approach was tested on in vivo data using 300 infant 3-D MR brain scans, which were qualitatively validated by an MR expert. In addition, it was quantitatively validated using 30 datasets based on three metrics: the Dice coefficient, the 95% modified Hausdorff distance, and absolute brain volume difference. Results showed the capability of the proposed approach, outperforming four widely used BETs: BET, BET2, brain surface extractor, and infant brain extraction and analysis toolbox. Experiments conducted also proved that the proposed framework can be generalized to adult brain extraction as well.

3D shape analysis of the brain cortex with application to dyslexia

To discriminate more accurately between dyslexic and normal brains, we detect the brain cortex variability through a spherical harmonic analysis that represents a 3D surface supported by the unit sphere, having a linear combination of special basis functions, called spherical harmonics (SHs). The proposed 3D shape analysis is carried out in five steps: (i) 3D brain cortex segmentation, with a deformable 3D boundary, controlled by two probabilistic visual appearance models (the learned prior and the estimated current appearance one); (ii) 3D Delaunay triangulation to construct a 3D mesh model of the brain cortex surface; (iii) mapping this model to the unit sphere; (iv) computing the SHs for the surface, and (v) determining the number of the SHs to delineate the brain cortex. We describe the brain shape complexity with a new shape index, the estimated number of the SHs, and use it for the K-nearest classification into the normal and dyslexic brains. Initial experiments suggest that our shape index is a promising supplement to the current dyslexia diagnostic techniques.

Shape-Based Detection of Cortex Variability for More Accurate Discrimination Between Autistic and Normal Brains

Autism is a complex developmental disability that typically appears during the first 3 years of life, and is the result of a neurological disorder that affects the normal functioning of the brain, impacting development in the areas of social interaction and communication skills. Early detection allows for treatments to be attempted, thus minimizing the impact of the autism on the individual. Given currently available diagnostic instruments, autism and other pervasive developmental disorders are difficult to detect in very young children. While shape based statistical analysis methods for autism are still in their early stages, current results show positive outlooks on the ability to detect differences between autistic and non-autistic patients. A framework is proposed that is capable of taking two-dimensional images from a standard medical scanner, and be able to construct a three-dimensional representation of the object and examine it through combination of its weighted linear spherical harmonics. The desired outcome is that a distinction can be made between the analysis of autistic and non-autistic brain data. The reconstruction analysis process involves linearly combining spherical harmonics of the corresponding mesh. It was expected that due to the complexity of the brain of an autistic subject it would require more iterations of reconstruction to reach convergence of the same error level as compared to the brain of a non-autistic subject. This was confirmed by the data. Using this method of analyzing the data a significant difference can be demonstrated between groups of examined subjects. The research clearly demonstrates that the non-autistic subjects’ data converges both faster and with a lower rate of error level than the data taken from a person with autism.

A new framework for automated identification of pathological tissues in contrast enhanced cardiac magnetic resonance images

A novel automated framework for quantification of myocardial viability in contrast enhanced cardiac magnetic resonance images (CE-CMRI) is proposed. The framework consists of three main steps. First, the inner and outer borders of the left ventricle (LV) wall (myocardium wall) are segmented from the surrounding tissue. Second, the pathological tissue in the myocardium wall is identified using a MAP-based classifier based on the visual appearance and spatial interaction of the LV pathological tissue as well as healthy tissue. Third, the myocardial viability is assessed and quantified based on measuring two parameters: the percentage of pathological tissue with respect to the area of the myocardium wall and the transmural extent of the pathological tissue in the myocardium wall. The transmural extent is estimated based on a new Partial Differential Equation (PDE) approach to determine point-to-point correspondences between the inner and outer borders of the pathological area as well as the myocardium wall. The proposed framework was tested on in-vivo CE-CMR images and validated with manual expert delineations of pathological tissue. Experiments and comparison results on real CE-CMR images confirm the robustness and accuracy of the proposed framework over the existing ones.

Atlas-based approach for the segmentation of infant DTI MR brain images

In this paper, we propose a new adaptive atlas-based technique for the automated segmentation of brain tissues (white matter and grey matter) from infant diffusion tensor images (DTI). Brain images and desired region maps (brain, Cerebrospinal fluid, etc.) are modeled by a joint Markov-Gibbs random field (MGRF) model of independent image signals and interdependent region labels. The proposed joint MGRF model accounts for the following three descriptors: (i) a 1st-order visual appearance to describe the empirical distribution of six features that has been estimated from the DTI in addition to the non-diffusion (b0) scans, (ii) 3D probabilistic atlases, and (iii) a 3D spatially invariant 2nd-order homogeneity descriptor. The 1st-order visual appearance descriptor, assuming each of the estimated DTI parameters are independent, is precisely approximated using our previously developed linear combination of discrete Gaussians (LCDG) intensity model that includes positive and negative Gaussian components. The 3D probabilistic atlases are learned using a subset of the 3D co-aligned training DTI brain images. The 2nd-order homogeneity descriptor is modeled by a 2nd-order translation and rotation invariant MGRF of region labels, with analytically estimated potentials. We tested our approach on 25 DTI brain images, and evaluated the performance on 5 manually segmented 3D DTI brain images to confirm the high accuracy of the proposed approach, as evidenced by the Dice similarity, Hausdorff distance, and absolute volume difference metrics.