Karel Bartusek

Soft-tissues image processing: comparison of traditional segmentation methods with 2D active contour methods

Soft-tissues Image Processing: Comparison of Traditional Segmentation Methods with 2D active Contour Methods The paper deals with modern methods of image processing, especially image segmentation, classification and evaluation of parameters. It focuses primarily on processing medical images of soft tissues obtained by magnetic resonance tomography (MR). It is easy to describe edges of the sought objects using segmented images. The edges found can be useful for further processing of monitored object such as calculating the perimeter, surface and volume evaluation or even three-dimensional shape reconstruction. The proposed solutions can be used for the classification of healthy/unhealthy tissues in MR or other imaging. Application examples of the proposed segmentation methods are shown. Research in the area of image segmentation focuses on methods based on solving partial differential equations. This is a modern method for image processing, often called the active contour method. It is of great advantage in the segmentation of real images degraded by noise with fuzzy edges and transitions between objects. In the paper, results of the segmentation of medical images by the active contour method are compared with results of the segmentation by other existing methods. Experimental applications which demonstrate the very good properties of the active contour method are given.

Magnetic field mapping around metal implants using an asymmetric spin-echo MRI sequence

Magnetic susceptibility of materials that are used for body implants causes much distortion in MR images. This paper deals with mapping the magnetic field induction in the vicinity of these materials. A modified spin-echo method is used, in which via inserting a time interval of a defined length the range of phase modulation is reduced below the value 2π. Even with a large B0 inhomogeneity it is not necessary to unwrap the phase jumps. The method is suitable for studying the effects of body-implant materials on MR images and for measuring their magnetic susceptibility.

Magnetic susceptibility measurement using 2D magnetic resonance imaging

The authors describe a nuclear magnetic resonance (NMR) method for measuring the magnetic susceptibility of bar-shaped samples that have an arbitrary cross-section and do not produce an MR signal. The method is based on the measurement of the 2D map of the reaction field in the vicinity of a non-ferromagnetic sample and on the calculation of magnetic susceptibility from a known reaction field. The verification of the technique was realized via modelling the measured configuration in the ANSYS program by means of the finite element method and through an experimental measurement of MR-compatible and MR-incompatible materials carried out on a tomograph. A great advantage of the proposed susceptibility evaluation method consists in the use of only standard commercially used devices without the need of any special sequences. The method is suitable for bar-shaped samples having an arbitrary cross-section; moreover, conditions are given for the selection of the cross-section/length ratio of a sample to be measured.

An NMR MULTIFID method for measurement of magnetic field gradient

The described MULTIFID method, that allows measurement of the time characteristic of the magnetic field gradient Gz(t), is based on the measurement of the instantaneous frequency of FID signals periodically recorded immediately after switching off the gradient pulse. The method was used for a fast and sufficiently accurate determination of the pre-emphasis time and amplitude constants. The preemphasis filter used allows one to shorten the decay time of the gradient magnetic field Gz from 200 ms to 100 mu s, and, thereby, to carry out a good compensation for the influence of eddy currents for the purposes of NMR spectroscopy and microscopy.

Mapping of magnetic field around small coils using the magnetic resonance method

The paper deals with measuring magnetic induction by imaging techniques based on magnetic resonance (MR). It describes experimental results of mapping the magnetic induction of helical and toroidal coils performed by modified MR imaging techniques. The results are compared with the theoretical calculation and with the measurement carried out using an ordinary magnetometer. The conclusions derived from the measurement will be used in the design of a new type of coil with complex electronic control of the magnetic field.

Automatic detection of brain tumors in MR images

This paper deals with automatic brain tumor detection in magnetic resonance images. The goal is to determine whether the MR image of a brain contains a tumor. The proposed method works with T2-weighted magnetic resonance images, where the head is vertically aligned. The detection is based on checking the left-right symmetry of the brain, which is the assumption for healthy brain. The algorithm was tested by fivefold cross-validation technique on 72 images of brain containing tumors and 131 images of healthy brain. The proposed method reaches the true positive rate of 91.16% and the true negative rate of 94.68%.

Wavelet-based de-noising techniques in MRI

The paper deals with techniques for the enhancement of magnetic resonance (MR) images using the wavelet analysis, which is assessed from the viewpoint of choosing the mother wavelet and the thresholding technique. Three parameters are used as objective criteria of the quality of image enhancement: the signal-to-noise ratio (SNR), image contrast, and linear approximation of edge steepness. Unlike most of the standard methods, which work exclusively with image magnitude, we also examined the influence of image phase, i.e. the image is processed as a complex signal. In addition to the interpretation of results, a short summary is given that deals with the choice of the optimal mother wavelet and thresholding technique for different types of MR images.

MR measurement technique of rapidly switched gradient magnetic fields in MR tomography

Gradient eddy currents, induced in the surrounding conductive structures in a magnetic resonance (MR) magnet, are a major problem in MR imaging, in localized MR spectroscopy and in many other MR experiments. We present a comparison of three methods of measuring the gradient time characteristics and the time changes of basic magnetic fieldB0 after the gradient is switched off. The methods are based on the selective excitation of a thin layer of the sample and on acquiring the MR signal obtained after the end of the gradient pulse and on the computation of the instantaneous frequency of the signal. At this point, the time gradient characteristic is proportional to the instantaneous frequency of the MR signal, which has a small signal-to-noise ratio. We use the characteristics measured to set the pre-emphasis parameters in a 200 MHz/200 mm MR scanner. From the results obtained by measurement it follows that all methods are convenient for simple and quick characterization of magnetic field gradient in MR tomographic magnets.

Diffusion MRI: mitigation of magnetic field inhomogeneities

Abstract The article reports on certain artifacts that emerge during the in vitro diffusion-weighted imaging of physical samples. In this context, the authors analyze the influence of magnetic field inhomogeneity, temperature, or eddy currents and consider artifact mitigation procedures. A technique reducing the examined spurious effects was designed, experimentally verified, and denominated as the three measurement method. The technique proved to be useful mainly for the evaluation of a DWI image measured with a diffusion gradient in the z axis, where the relative measurement error decreased to 3.38 % (during measurement using two images, the relative error was greater than 19 %). For small errors within the measurement of diffusion constants of a deionized water sample (< 5 %) it was necessary to select a b-factor value larger than 200·106 s.m-2. Temperature stabilization with accuracy better than 0.1 °C during the entire measuring process is a necessary prerequisite for the measurement of biological or material samples with relative accuracy lower than 1 %.

Determination of pre-emphasis constants for eddy current reduction

Magnetic field gradients play a fundamental role in fast magnetic resonance imaging (MRI) methods and spectroscopy imaging. Precise information on gradient waveform shape and rise times is often very useful in experimental MRI. We present a simple method for the measurement of the gradient time course and static magnetic field changes. The method does not require any specialized hardware and can be used with a standard volume coil and a special phantom filled with deionized or doped water. The method is based on an analysis of the instantaneous frequency variation of an MR signal in the time domain, acquired from a mechanically selected thin slice of phantom located at the gradient isocentre. The measurement and approximation of the course of time gradients and of the changes in the static magnetic field result in pre-emphasis constants that substitute the constants determined manually in a time-consuming manner. The described method facilitates the determination of pre-emphasis constants for the compensation of eddy current effects in MR systems. Our results show that residual gradients in the region of interest within 0.5 ms after the gradient is switched off are smaller than 5 µT m−1 (for an applied test gradient of 72 mT m−1) and the shift of the static field is smaller than 0.2 µT. This accuracy is very important for the development of MR spectroscopic imaging technologies using samples with very short relaxation times, and for EPI pulse sequences.