Žiga Špiclin

Image registration for visual inspection of imprinted pharmaceutical tablets

Image registration plays a vital role in visual quality inspection of tablets with imprints. In this paper, three registration methods, based on direct matching of pixel intensities, principal axes matching and circular profile matching, were proposed and objectively evaluated on real, large and representative image database (1,634 images) of various types of non-defective and defective tablets. Comparing against a devised “gold standard” registration and considering execution times and different registration tasks, circular profile matching method proved to be a powerful image processing tool for improving visual quality inspection of imprinted tablets in real time.

Geometric calibration of a hyperspectral imaging system

Every imaging system requires a geometric calibration to yield accurate optical measurements. Geometric calibration typically involves imaging of a known calibration object and finding the parameters of a camera model and a model of optical aberrations. Optical aberrations can vary significantly across the wide spectral ranges of hyperspectral imaging systems, which can lead to inaccurate geometric calibrations if conventional methods were used. We propose a method based on a B-spline transformation field to align the spectral images of the calibration object to the model image of the calibration object. The degree of spatial alignment between the ideal and the spectral images is measured by normalized cross correlation. Geometric calibration was performed on a hyperspectral imaging system based on an acousto-optic tunable filter designed for the near-infrared spectral range (1.0-1.7microm). The proposed method can accurately characterize wavelength dependent optical aberrations and produce transformations for efficient subpixel geometric calibration.

Stratified mixture modeling for segmentation of white-matter lesions in brain MR images

Accurate characterization of white-matter lesions from magnetic resonance (MR) images has increasing importance for diagnosis and management of treatment of certain neurological diseases, and can be performed in an objective and effective way by automated lesion segmentation. This usually involves modeling the whole-brain MR intensity distribution, however, capturing various sources of MR intensity variability and lesion heterogeneity results in highly complex whole-brain MR intensity models, thus their robust estimation on a large set of MR images presents a huge challenge. We propose a novel approach employing stratified mixture modeling, where the main premise is that the otherwise complex whole-brain model can be reduced to a tractable parametric form in small brain subregions. We show on MR images of multiple sclerosis (MS) patients with different lesion loads that robust estimators enable accurate mixture modeling of MR intensity in small brain subregions even in the presence of lesions. Recombination of the mixture models across strata provided an accurate whole-brain MR intensity model. Increasing the number of subregions and, thereby, the model complexity, consistently improved the accuracy of whole-brain MR intensity modeling and segmentation of normal structures. The proposed approach was incorporated into three unsupervised lesion segmentation methods and, compared to original and three other state-of-the-art methods, the proposed modeling approach significantly improved lesion segmentation according to increased Dice similarity indices and lower number of false positives on real MR images of 30 patients with MS.

Combining Unsupervised and Supervised Methods for Lesion Segmentation

White-matter lesions are associated to several diseases, which can be characterized by neuroimaging biomarkers through lesion segmentation in MR images. We present a novel automated lesion segmentation method consisting of an unsupervised mixture model based extraction of candidate lesion voxels, which are subsequently classified by a random decision forest (RDF) using simple visual features like multi-sequence MR intensities sourced from connected voxel neighborhoods. The candidate lesion extraction prior to RDF training and classification balanced the number of non-lesion and lesion voxels and the number of non-lesion classes versus a lesion class. Thereby, the RDF established highly discriminating decision rules based on such simple visual features, which have the benefit of no computational overhead and easy extraction from the MR images. On MR images of 18 patients with multiple sclerosis the proposed method achieved the median Dice similarity of 0.73, sensitivity of 0.90 and positive predictive value of 0.61, which indicate accurate segmentation of white-matter lesions.

Computer-Aided Detection and Quantification of Intracranial Aneurysms

Early detection, assessment and treatment of intracranial aneurysms is important to prevent rupture, which may cause death. We propose a framework for detection and quantification of morphology of the aneurysms. A novel detector using decision forests, which employ responses of blobness and vesselness filters encoded in rotation invariant and scale normalized frequency components of spherical harmonics representation is proposed. Aneurysm location is used to seed growcut segmentation, followed by improved neck extraction based on intravascular ray-casting and robust closed-curve fit to the segmentation. Aneurysm segmentation and neck curve are used to compute three morphologic metrics: neck width, dome height and aspect ratio. The proposed framework was evaluated on ten cerebral 3D-DSA images containing saccular aneurysms. Sensitivity of aneurysm detection was 100% at 0.4 false positives per image. Compared to measurements of two expert raters, the values of metrics obtained by the proposed framework were accurate and, thus, suitable for assessing the risk of rupture.

Method for 3D-2D Registration of Vascular Images: Application to 3D Contrast Agent Flow Visualization

Endovascular image guided interventions involve catheter navigation through the vasculature to the treatment site under guidance of live 2D projection images. During treatment materials are delivered through the catheter that requires information about the blood flow direction, obtained by injecting contrast agent and observing its propagation on the live 2D images. To facilitate navigation and treatment the information from the live 2D images can be superimposed on a 3D vessel tree model, extracted from pre-interventional 3D images. However, the 3D and live 2D images first need to be spatially corresponded by a 3D-2D registration. In this paper, we propose a novel 3D-2D registration method based on matching orientations of 3D vessels’ centerlines to the edges of live 2D images. Results indicate that the proposed 3D-2D registration is highly robust and feasible for real-time execution (<1 s). Example of 3D contrast flow visualization also demonstrates the potential for real clinical application.

Validation of White-Matter Lesion Change Detection Methods on a Novel Publicly Available MRI Image Database

Changes of white-matter lesions (WMLs) are good predictors of the progression of neurodegenerative diseases like multiple sclerosis (MS). Based on longitudinal magnetic resonance (MR) imaging the changes can be monitored, while the need for their accurate and reliable quantification led to the development of several automated MR image analysis methods. However, an objective comparison of the methods is difficult, because publicly unavailable validation datasets with ground truth and different sets of performance metrics were used. In this study, we acquired longitudinal MR datasets of 20 MS patients, in which brain regions were extracted, spatially aligned and intensity normalized. Two expert raters then delineated and jointly revised the WML changes on subtracted baseline and follow-up MR images to obtain ground truth WML segmentations. The main contribution of this paper is an objective, quantitative and systematic evaluation of two unsupervised and one supervised intensity based change detection method on the publicly available datasets with ground truth segmentations, using common pre- and post-processing steps and common evaluation metrics. Besides, different combinations of the two main steps of the studied change detection methods, i.e. dissimilarity map construction and its segmentation, were tested to identify the best performing combination.

Fast and Robust 3D to 2D Image Registration by Backprojection of Gradient Covariances

Visualization and analysis of intra-operative images in imageguided radiotherapy and surgery are mainly limited to 2D X-ray imaging, which could be beneficially fused with information-rich pre-operative 3D image information by means of 3D-2D image registration. To keep the radiation dose delivered by the X-ray system low, the intra-operative imaging is usually limited to a single projection view. Registration of 3D to a single 2D image is a very challenging registration task for most of current state-of-the-art 3D-2D image registration methods. We propose a novel 3D-2D rigid registration method based on evaluation of similarity between corresponding 3D and 2D gradient covariances, which are mapped into the same space using backprojection. Normalized scalar product of covariances is computed as similarity measure. Performance of the proposed and state-of-the-art 3D-2D image registration methods was evaluated on two publicly available image datasets, one of cerebral angiograms and the other of a spine cadaver, using standardized evaluation methodology. Results showed that the proposed method outperformed the current state-of-the-art methods and achieved registration accuracy of 0.5 mm, capture range of 9 mm and success rate >80%. Considering also that GPU-enabled execution times ranged from 0.5-2.0 seconds, the proposed method has the potential to enhance with 3D information the visualization and analysis of intra-operative 2D images.

Automated segmentation of MS lesions in brain MR images using localized trimmed-likelihood estimation

Diagnosis and prognosis of patients with multiple sclerosis (MS) rely on quantitative markers derived from the analysis of magnetic resonance (MR) images. To compute these markers, a segmentation of lesions in the brain tissues, which are characteristic for MS disease, is needed. In this paper, we propose an unsupervised method for segmenting MS lesions that employs localized trimmed-likelihood estimation (TLE) to model the intensity distributions of normal appearing brain tissues (NABT). Compared to the original whole-brain TLE approach, the proposed method employs a set of three-component Gaussian mixture models for each of the spatially localized and non-overlapping subregions of the brain. The subregions were assigned by the customized balanced box decomposition that takes into account the spatial distribution and the cardinality of NABT tissues, as obtained from the initial whole-brain TLE. The proposed method was tested and compared to the original TLE approach on publicly available synthetic BrainWeb datasets. The results indicate a higher average Dice similarity coefficient both for the segmentation of NABT and MS lesions by using the proposed spatially localized TLE as compared to the original whole-brain TLE, which is due to the fact that the proposed method yields a more accurate NABT model and thus detects fewer false NABT outliers.

Correction of axial optical aberrations in hyperspectral imaging systems

In hyper-spectral imaging systems with a wide spectral range, axial optical aberrations may lead to a significant blurring of image intensities in certain parts of the spectral range. Axial optical aberrations arise from the indexof- refraction variations that is dependent on the wavelength of incident light. To correct axial optical aberrations the point-spread function (PSF) of the image acquisition system needs to be identified. We proposed a multiframe joint blur identification and image restoration method that maximizes the likelihood of local image energy distributions between spectral images. Gaussian mixture model based density estimate provides a link between corresponding spatial information shared among spectral images so as to find and restore the image edges via a PSF update. Model of the PSF was assumed to be a linear combination of Gaussian functions, therefore the blur identification process had to find only the corresponding scalar weights of each Gaussian function. Using the identified PSF, image restoration was performed by the iterative Richardson-Lucy algorithm. Experiments were conducted on four different biological samples using a hyper-spectral imaging system based on acousto-optic tunable filter in the visible spectral range (0.55 - 1.0 μm). By running the proposed method, the quality of raw spectral images was substantially improved. Image quality improvements were quantified by a measure of contrast and demonstrate the potential of the proposed method for the correction of axial optical aberrations.