Andrius Usinskas

Erratum to: Optimal differentiation of high- and low-grade glioma and metastasis: a meta-analysis of perfusion, diffusion, and spectroscopy metrics.

Introduction To perform a meta-analysis of advanced magnetic resonance imaging (MRI) metrics, including relative cerebral blood volume (rCBV), normalized apparent diffusion coefficient (nADC), and spectroscopy ratios choline/creatine (Cho/Cr) and choline/N-acetyl aspartate (Cho/NAA), for the differentiation of high- and low-grade gliomas (HGG, LGG) and metastases (MTS).

Automatic Ischemic Stroke Segmentation Using Various Techniques

Different methods of automatic segmentation of human brain ischemic stroke area in the computerized tomography scans are compared. Experts-radiologists performed the evaluation of segmentation techniques. A methodology of qualitative evaluation of the investigated methods is proposed. The best viability showed histogram, gray level co-occurrence matrix, mean and standard deviation methods, and supervised artificial neural network technique.

Automatic contouring of segmented human brain ischemic stroke region on ct images 1

Abstract Investigation of the relationship between stroke area size and patient status implies evaluation of the stroke area location, volume and shape. Computed tomography (CT) examination became very popular in such type of investigations due to its moderate price and less discomfort for patient in comparison with other techniques. We propose an algorithm for segmented CT images post‐processing, which consists of several stages: filtering of CT slices, smoothing and extension of stroke region boundary, filling of stroke space, and computing of stroke volume via all slices. Post‐processing of several CT images using this technique showed that all accidental points can be filtered successfully and therefore the aim of ischemie stroke area determination can be reached. We are convinced that the quality of initial information (results of stroke area segmentation) plays a crucial role for later image processing.

Improvements on the Gray Level Co-occurrence Matrix Technique to Compute Ischemic Stroke Volume

The purpose of this work was to apply and test Haralick’s gray level co-occurrence matrix (GLCM) technique for automatic calculation and segmentation of the ischemic stroke volume from CT images. For this task, the 3-nearest neighbors classifier was trained to perform stroke and non-stroke area classification. The segmentation and classification results were compared versus a manual segmentation. Approximately half of the automatically computed and segmented stroke volumes from CT images differed less than 15 % from the corresponding manually segmented stroke volumes.

Generating of reformat slices in neural and cardio‐tomography

Abstract A lot of medical diagnostics problems are related to the reconstruction of three dimension (3D) images from the cross sections of the regions of interest. Such reconstructions are very desirable for evaluation of disease or planning of surgical treatment. This paper reviews recent 3D preprocessing work of authors in human brain blood vessels structure recognition and localization of aneurysms as well as analysis of the right ventricular of the human heart. Here we present some approximation techniques for generation of reformatted images.

EMBEDDED DSP INTENSIVE PROJECT 2010

In this paper, we describe the first Embedded DSP Intensive Project (eDSP IP) held on August 2010 in Helsinki Metropolia University of Applied Sciences. The general idea was to bring together teachers from four European University to integrate their high expertise on different electronics and IT engineering fields, thus creating and delivering a series of multidisciplinary lectures. This intensive project was supported by the funds of the Erasmus Intensive Programme of the European Commission.

Computing Volume of the Heart's Right Ventricle Using 2D Echocardiography Images

AbstractThis paper presents a new way for computing the volume of the hearts right ventricle. The technique involves manual marking of three critical contours of the ventricles regions on 2D echocardiography images. The contours are then deformed semi-automatically to keep them orthogonal and concurrent in four-chamber plane, transverse (short-axis) plane, and lateral plane. The ventricles surface is generated and its volume computed automatically. The proposed model was validated by comparing its estimate with actual measurements using 3D echocardiography imaging.

Modelling of the MRI perfusion process of human head

In the paper two models of the brain perfusion for dynamic susceptibility contrast magnetic resonance imaging technique are introduced. Estimation of the cerebral perfusion parameters is done by two approaches. Both approaches are based on non-linear regression method. Parametric maps of perfusion parameters are given and compared. In conclusion it is resumed what factors can effect the final results.

Extraction of centre line from curvilinear objects

This tutorial paper surveys three methods designed to detect the centre line in curvilinear structures. These are the iterative thinning, Steger’s and derivatives’ methods. We aim to illustrate the effectivness of the chosen methods for processing laser-trace images and magnetic resonance slices of human head. The essence of each three methods is presented and important parameters are discussed. Experiments have been carried out and results are discussed in the light of the quality of the centre line produced and work time of all three methods.