Ultrasound computed tomography in breast imaging: first clinical results of a custom-made scanner

PURPOSE

To test a system using ultrasound computed tomography (USCT) that superimposes ultrasound data acquired in one cross-sectional plane from multiple angles around the breast (Full Angle Spatial Compounding, FASC) and to reconstruct the distribution of the speed of sound in the breast (SoS reconstruction).

MATERIALS AND METHODS

We developed a system combining a conventional ultrasound scanner with a PC-controlled mechanical setup integrated in a custom-made examination couch. In a feasibility study, 3 volunteers (age 26 - 74 years) and one patient with breast cancer were studied. Subjects were placed in the prone position on this couch, with the breast hanging in a water tank. The ultrasound probe was moved in several planes around the breast. A curved reflector that followed the movement of the probe behind the breast was used to calculate the SoS within the breast tissue. Echo-data was processed offline by custom-made software to calculate both FASC and SoS images.

RESULTS

In FASC images a reduction of artifacts (i. e. shadowing of Cooper's ligaments and irregular edges of inhomogeneous lesions) and speckles as well as clear visualization of the inner architecture of the breast was achieved. SoS images delivered further diagnostic information and helped to compensate for geometric distortions in the computed images. Difficulties in the visualization of lesions near the thoracic wall and/or the axillary are limitations of this technique.

CONCLUSION

The first clinical results of USCT imaging have proven its feasibility as an automated and standardized technique for breast imaging.

Ultrasonic microscanning

A detailed review is given of the application of high-frequency ultrasound (HFUS) at frequencies of 20 MHz and above for high-resolution, cross-sectional imaging of biological soft tissue. The state of the art of HFUS imaging systems is discussed with respect to the underlying engineering concepts, system designs, and available transducer technology. Furthermore, the dependency of the spatial resolution on the system's parameters is analysed. Skin imaging, eye imaging, small animal imaging for preclinical research, and intravascular ultrasound in coronary arteries for arteriosclerotic disease diagnostics are presented as examples for the application of HFUS imaging in medical diagnostics. It is shown that, in the frame of the indicated applications, ultrasound in the frequency range 20-100MHz gives a good compromise between the contrary demands for a good spatial resolution and a sufficiently large penetration depth of ultrasound waves into the tissue. Scanning schemes for the imaging of tissue morphology are considered, including spatial compounding as a multidirectional imaging technique.

Sonohistology - ultrasonic tissue characterization for prostate cancer diagnostics

A computer-aided diagnostic system for imaging prostate cancer has been developed in order to supplement today's conventional methods for the early detection of prostate carcinoma. The system is based on analysis of the spectral content of radiofrequency ultrasonic echo data in combination with evaluations of textural, contextual, morphological and clinical features in a multiparameter approach. A state-of-the-art, non-linear classifier, the so-called adaptive network-based fuzzy inference system, is used for higher-order classification of the underlying tissue-describing parameters. The system has been evaluated on radio-frequency ultrasound data originating from 100 patients using histological specimens obtained after prostatectomy as the gold standard. Leave-one-out cross-validation over patient data sets results in areas under the ROC curve of 0.86 +/- 0.01 for hypoechoic and hyperechoic tumors and of 0.84 +/- 0.02 for isoechoic tumors, respectively.

Palpation imaging using a haptic system for virtual reality applications in medicine

In the field of medical diagnosis, there is a strong need to determine mechanical properties of biological tissue, which are of histological and pathological relevance. Malignant tumors are significantly stiffer than surrounding healthy tissue. One of the established diagnosis procedures is the palpation of body organs and tissue. Palpation is used to measure swelling, detect bone fracture, find and measure pulse, or to locate changes in the pathological state of tissue and organs. Current medical practice routinely uses sophisticated diagnostic tests through magnetic resonance imaging (MRI), computed tomography (CT) and ultrasound (US) imaging. However, they cannot provide direct measure of tissue elasticity. Last year we presented the concept of the first haptic sensor actuator system to visualize and reconstruct mechanical properties of tissue using ultrasonic elastography and a haptic display with electrorheological fluids. We developed a real time strain imaging system for tumor diagnosis. It allows biopsies simultaneously to conventional ultrasound B-Mode and strain imaging investigations. We deduce the relative mechanical properties by using finite element simulations and numerical solution models solving the inverse problem. Various modifications on the haptic sensor actuator system have been investigated. This haptic system has the potential of inducing real time substantial forces, using a compact lightweight mechanism which can be applied to numerous areas including intraoperative navigation, telemedicine, teaching and telecommunication.

Vibrography: first experimental results in swine brains

OBJECT

The aim of this study was to determine whether vibrography, an ultrasound-based real-time strain imaging method for registering the elastic properties of tissue, is superior to conventional ultrasound imaging techniques for detecting low-contrast space-occupying lesions in brain tissue and for delineating the boundaries between such lesions and the surrounding tissue.

METHODS

As our experimental model we used swine brains taken from freshly slaughtered pigs. After injecting agarose into these brains at different depths, we compared both the conventional ultrasonographic images and the elastographic images of the region of interest with the corresponding anatomical brain sections.

RESULTS

In 83.6 % of the experiments, it was possible to detect the polymerized agarose in the brain tissue with vibrographic techniques. In 17 experiments agarose lesions which were not detectable by ultrasound were visualized via vibrography. Furthermore, statistical analysis revealed that elastography is a more precise tool than conventional ultrasound for determining lesion size.

CONCLUSION

These findings indicate that vibrography is a promising real-time imaging method with numerous potential applications in the field of neurosurgery. Visualization of the elastic properties provides the neurosurgeon with additional data on the lesion and the boundary between the lesion and the surrounding tissue.

Geometrical optimization of a phased array coil for high-resolution MR imaging of the carotid arteries

The geometry of an RF phased-array receiving coil for high-resolution MRI of the carotid artery, particularly the bifurcation, was optimized with respect to signal-to-noise ratio (SNR). A simulation tool was developed to determine homogeneity, sensitivity, and SNR for a given imaging situation. The algorithm takes into account the coil geometry, the parameters of the measured object, and the imaging parameters of the pulse sequence. The coil with the optimum geometry was implemented as a receive-only coil for 1.5 T and comparative SNR measurements with different coils were performed. The experimental SNR measurements verified the simulations. The optimized carotid artery phased array offered the best SNR over the desired field of view. In vivo high-resolution MRI of the carotid arteries of healthy volunteers and patients with known stenosis was conducted with the optimized phased array coil. The capability of the phased array coil for resolving components within the carotid artery walls is demonstrated. Magn Reson Med 50:439-443, 2003.

[Prostate cancer diagnosis using ultrasound elastography. Introduction of a novel technique and first clinical results]

During the last decade screening has improved prostate cancer detection. The main reason for this development is a better understanding of the margins of prostate-specific antigen (PSA) serum levels and the classification of PSA subtypes. In contrast, the introduction of transrectal ultrasound has not led to a measurable change in the prostate cancer detection rate. Our aim was to develop a novel ultrasound system for the acquisition of elastographic images of the prostate and evaluate the system regarding its clinical applicability. We used a technically modified conventional ultrasound system and analyzed the high-frequency ultrasonic data with a computer program. The first patient-based results suggest that elastography allows an accurate measurement of tumor size and localization in contrast to conventional transrectal ultrasound. Elastography visualizes different tissue elasticities to distinguish benign and cancerous tissue. Thus, we were able to even correctly classify prostate cancer lesions which are iso- or hyperechoic in B-mode sonography.

Strain imaging with intravascular ultrasound array scanners: validation with phantom experiments

Intravascular Ultrasound (IVUS) is routinely used in interventional cardiology for imaging coronary plaque morphology. However, the use of B-mode images for tissue characterization and detection of vulnerable coronary plaques is limited. Strain imaging with ultrasound is a new modality that provides additional information for tissue characterization by imaging differences in tissue stiffness. The aim is to differentiate between vulnerable (soft) plaques and less dangerous calcified (hard) plaques. In this work, the applicability of a time efficient strain imaging algorithm in conjunction with data from IVUS array transducers is evaluated. Unfocused radiofrequency (rf) data from the transducer array is acquired using custom made hardware. Rf line reconstruction is performed offline by synthetic aperture focusing techniques. Vessel mimicking phantoms of different geometries and material stiffness are made from agar and Polyvinyl Alcohol Cryogel (PVA). Experiments are conducted in a water tank and a water column is used for applying intraluminal pressure differences required for strain imaging. The results show that strain images can be calculated with A-lines reconstructed from unfocused rf raw data. Regions of different stiffness can be identified qualitatively by local strain variations. With the used algorithm strains of up to 2% can be imaged without significant decor-relation.

Real-time detection of vessel diameters with ultrasound

Transcutaneous vessel imaging is a frequently used ultrasound imaging modality in medicine. The measurement of vessel diameters can be done with conventional B-mode imaging systems, which work at frame rates up to 100 Hz. Furthermore, there are special systems available, which can track vessel walls very precisely using the phase of signals that are sent at frame rates up to several thousand Hz. Though, such systems are usually not able to provide the examiner with 2D images of the object. With respect to brachial artery flow-mediated vasodilatation (FMD), which is frequently used as a measure of endothelial function, it is necessary to observe diameter changes of small arterial vessels noninvasively for several minutes at a high resolution. In the past, the diameter had to be measured manually in tedious postprocessing of ECG-gated image sequences. We developed a system composed of a Siemens Omnia ultrasound system with a VF13-5 transducer (9 MHz center frequency) and a personal computer, that is capable of calculating vessel diameter changes with an accuracy below the wavelength of the ultrasound system in real-time at a frame rate of 27 Hz. We implemented a two-dimensional active contour model using the Viter-bi-algorithm and a phase-sensitive vessel wall tracking algorithm, in order to guarantee both, geometric information and accuracy. Results from carotid and brachial arteries show that arterial pulsations below 0.1 mm can be visualized reliably over several minutes. With this system we want to find out, if FMD is suitable for an individual assessment of the risk for cardiovascular diseases.

In vivo biomicroscopy of the skin with high-resolution magnetic resonance imaging and high frequency ultrasound

Noninvasive imaging and characterization of the skin is of great interest in dermatology. In order to get relevant diagnostic information, high-resolution imaging techniques have to be applied. Ultrasonic imaging is a potential method for this purpose where the special requirements concerning the spatial resolution make it essential to apply high frequency ultrasound (HFUS). Alternatively, magnetic resonance imaging (MRI), being a very promising imaging modality, also shows the perspective of becoming a valuable diagnostic tool in dermatology. However, to account for the small dimensions of the structures under observation, very specialized system designs have to be developed. In this paper, a HFUS imaging system working in the 50 MHz and 100 MHz range is applied for high-resolution skin imaging. Furthermore, a commercial MRI-system was equipped with specially designed low noise rf (radio frequency) coils with minimized volume, and customized imaging sequences were applied to optimize the signal-to-noise ratio. With HFUS and high-resolution magnetic resonance (HR-MR) imaging complementary imaging techniques for in vivo biomicroscopy of the skin are available.

A nonuniform sampling approach for fast ultrasonic flow imaging

Conventional Pulsed Wave Doppler (PWD) systems acquire an ensemble of N echoes per beam line at a constant pulse repetition frequency fprf, so that the pulse repetition interval equals Tpri = 1/fpn. The total time span determines the velocity resolution, and Tpri the unambiguous velocity range. The ensemble size N is by approximation inversely proportional to the frame rate, assuming that the system performs interleaving. For a given frame rate, a tradeoff can only be made between velocity resolution and velocity range. We propose an approach that allows increasing velocity resolution or range while keeping the frame rate constant. The approach is based on nonuniform sampling, i.e. sampling with varying sampling intervals. Thus, for a given ensemble size N a larger total time span, which would increase velocity resolution, or a shorter minimal Tpn, which would increase the velocity range, may be chosen. The conventional Doppler signal processing techniques are not compatible with nonuniform sampling. We, therefore, developed a velocity estimation algorithm for arbitrary sampling that is based on cross correlation. Furthermore, an adaptive wall filter was implemented that differentiates between tissue motion and blood flow. The new approach was successfully tested with in vitro and in vivo data.

Comparison of high frequency ultrasound and optical coherence tomography as modalities for high resolution and non invasive skin imaging

High frequency ultrasound (HFUS) and optical coherence tomography (OCT) are techniques for high resolution imaging of tissues. The penetration depth of these modalities is limited, but it is sufficiently large enough for non invasive skin imaging. HFUS and OCT are based on the same concept. Waves (ultrasonic waves, respectively light waves) propagate along a narrow beam, are backscattered at tissue inhomogeneities and analyzed over time of flight to obtain spatially resolved morphological information. The objective of this paper is to compare HFUS and OCT in terms of resolution, dynamic range and contrast and to assess their value as tools for high resolution skin imaging. Measurements on phantoms and in vivo have been performed with a 100 MHz ultrasound system and an OCT-scanner working in the near infrared spectrum at 1300 nm wave-length. From the measurements, it can be concluded that OCT delivers an almost isotropic resolution (axial resolution about 5.8 microns, lateral resolution about 4.1 microns), whereas the resolution of the investigated HFUS system is more anisotropic (axial resolution about 9.3 microns, lateral resolution about 60 microns). HFUS and OCT show different penetration depths and a different contrast. Both techniques can, therefore, be combined advantageously in a multimodality approach to account for their individual characteristics.

Ultrasonic tissue characterization for prostate diagnostics: spectral parameters vs. texture parameters

An ultrasonic multi-feature tissue characterizing system for the detection of prostate cancer is presented. The system is based on the processing of radio frequency (RF) ultrasonic echo data. Data from 100 patients was acquired in a clinical study. Parameters are extracted from the RF echo data and classified using two adaptive network-based fuzzy inference systems (FIS) working in parallel as a nonlinear classifier. Next to spectral parameters, conventional texture parameters are calculated using demodulated and log-compressed echo data. In the first approach, the classifier is trained on both, spectral and texture parameters. In the second approach, the classifier is only trained on texture parameters. Classification results of both approaches are compared and it is demonstrated, that only the use of spectral parameters yields satisfying classification results. Results of a minimum distance classifier (MDC) are presented for comparison with the fuzzy inference system. For the final fuzzy inference systems used in this approach, the area under the ROC curve is between 84% and 86% for the combined approach and between 70% and 74% for the approach based on texture parameters only.

[Measurement of flow mediated vasodilation (FMD) using kalman-filtering]

Brachial artery flow-mediated vasodilation is increasingly used as a measure of endothelial function. High resolution ultrasound provides a noninvasive method to observe this flow-mediated vasodilation by monitoring the diameter of the artery over time. In the past, the diameter had to be measured in tedious postprocessing routines, usually by the examiner himself. We present a system, which is able to process ultrasound rf-data in realtime. On that system, we implemented a kalman filter, which makes the tracking of both vessel walls possible. The diameter can be calculated accurately, taking into account process noise as well as measurement noise.

[Model construction for reperfusion of the isolated pig liver]

In recent years thermal ablation of liver tumors as a minimally invasive method became a promising alternative to conventional strategies such as chemotherapy or resection of liver tissue. Thereby an electrode is placed inside the tumor delivering energy in the form of high frequency current into the target volume to achieve and maintain a tissue temperature between 60 and 100 degrees C. Cells exposed to this thermic stress undergo coagulation necrosis and are irreversibly damaged. To protect vital liver structures from heat, it is necessary to develop an online temperature monitoring system. An experimental setup perfusing isolated pig livers under physiological conditions with 0.9% NaCl solution was established to develop and evaluate the measuring technique.

The Ruhr Center of Competence for Medical Enginnering (Kompetenzzentrum Medizintechnik Ruhr KMR, Bochum)

The profile and the projects of the Ruhr-Center of Competence for Medical Engineering at the Ruhr-University Bochum (Germany) will be described. Main topic of the KMR is medical ultrasound with emphasis on image based tissue characterization including elastography and multimodality concepts, mainly in combination with other non-ionizing imaging modalities. Project aims are early detection of cancer (skin, prostate), vessel and perfusion diagnostics (early detection of arteriosclerosis, cardiac arteries, stroke), and intraoperative navigation using ultrasound.

[Ultrasound-based imaging modalities for thermal therapy monitoring]

Thermal therapy has been established as an alternative and minimally invasive approach for the treatment of tumors. During a thermal therapy tissue is heated locally up to above 60 degrees C. Cancerous tissue can thus be destroyed by coagulation. At present there are no suitable imaging modalities available for an accurate real-time monitoring of the coagulation process. A subproject of the Ruhr Center of Competence for Medical Engineering (KMR Bochum) aims at developing an ultrasound-based, real-time capable monitoring system for thermal therapy. Therefore several tissue characterizing imaging modalities will be combined in a new multimodal concept. Initial experiments with porcine liver in vitro have shown that real-time monitoring of a thermal therapy using various imaging methods simultaneously will be possible.

Registration of bone surfaces, extracted from CT-datasets, with 3D ultrasound

An essential task of computer assisted surgery is the registration of preoperative image data with the coordinate system of the operating room. This can be reached by using intraoperative imaging and registrating preoperative and intraoperative datasets. For intraoperative imaging ultrasound is a powerful tool due to the lack of ionizing radiation and because of its fast, inexpensive and easy data acquisition. We propose a surface volume matching algorithm for the registration of bone surfaces and ultrasound volume data. The bone surface is estimated from the preoperative CT data by taking into account that ultrasound only shows parts of the bone surface. By our method reliable matching results are obtained. They are shown with data of the lumbar spine.

A haptic sensor-actor-system based on ultrasound elastography and electrorheological fluids for virtual reality applications in medicine

Mechanical properties of biological tissue represent important diagnostic information and are of histological relevance (hard lesions, "nodes" in organs: tumors; calcifications in vessels: arteriosclerosis). The problem is, that such information is usually obtained by digital palpation only, which is limited with respect to sensitivity. It requires intuitive assessment and does not allow quantitative documentation. A suitable sensor is required for quantitative detection of mechanical tissue properties. On the other hand, there is also some need for a realistic mechanical display of those tissue properties. Suitable actuator arrays with high spatial resolution and real-time capabilities are required operating in a haptic sensor actuator system with different applications. The sensor system uses real time ultrasonic elastography whereas the tactile actuator is based on electrorheological fluids. Due to their small size the actuator array elements have to be manufactured by micro-mechanical production methods. In order to supply the actuator elements with individual high voltages a sophisticated switching and control concept have been designed. This haptic system has the potential of inducing real time substantial forces, using a compact lightweight mechanism which can be applied to numerous areas including intraoperative navigation, telemedicine, teaching, space and telecommunication.

Registration of 3D CT and ultrasound datasets of the spine using bone structures

OBJECTIVE

In navigated orthopedic surgery, accurate registration of bones is of major interest. Usually, this registration is performed using landmarks positioned directly on the bone surface. These landmarks must be exposed during surgery. Our goal is to avoid the exposure of bone surface for the sole purpose of registration by using an intraoperative ultrasound device that can localize the bone through tissue.

METHOD

We propose an algorithm for the registration of CT and ultrasound datasets that takes into account the fact that ultrasound produces very noisy images (speckle) and shows only parts of the bone surface. This part is made from the CT dataset. Next, a surface volume registration is performed by searching for a position of the estimated surface that maximizes the average gray value of the voxels in the ultrasound dataset covered by the surface.

RESULTS

The algorithm was implemented and validated using an ex vivo preparation of a human lumbar spine with surrounding muscle tissue. On the basis of this data, the method has a large radius of convergence and a repeatability of 0.5 mm for displacement and 0.5 degrees for rotation.

CONCLUSIONS

A robust algorithm for the registration of 3D CT and ultrasound datasets is presented. The computation time seems sufficiently short to permit intraoperative use.

Freehand ultrasound elastography of breast lesions: clinical results

We developed a freehand method for ultrasound elastography, which can be applied during a routine sonographic examination with off-line calculation of strain images (elastograms). Forty-eight patients with 53 breast lesions were examined and, after biopsy or operation, histologic reports were available for all lesions. The correlation coefficient of time delay estimates was used as a quality criterion for the subsequent calculation of elastograms. Beyond the qualitative evaluation of elastograms, we suggested a semiquantitative approach. For that purpose, the elastogram of each lesion was normalized to an overall strain of 1% (i.e., the average strain in the image was set to 1%). After normalization, we determined mean strain values inside and outside of each lesion, respectively. Defining solid lesions as benign and malignant lesions except for fibrous mastopathy, we found significant difference in strain between solid lesions and their surrounding tissue. However, that result must not be misunderstood to suggest that it was possible to distinguish benign from malignant lesions in general. Still, we address the potential of ultrasound elastography to improve the detection and localization of breast lesions as well as their differential diagnosis. Besides, we developed a freehand applicator for further studies, which guarantees a homogeneous axial compression regardless of the experience of the examiner.

Synthetic aperture-based beam compression for intravascular ultrasound imaging

In this paper, intravascular ultrasound (IVUS) images acquired with a 64-element array transducer using a multistatic acquisition scheme are presented. The images are reconstructed from a collection of pulse-echo measurements using a synthetic aperture array imaging technique. The main limitations of IVUS imaging are a poor lateral resolution and elevated grating lobes caused by the imaging geometry. We propose a Synthetic Aperture Focusing Technique (SAFT), which uses a limited number of A-scan signals. The focusing process, which is performed in the Fourier domain, requires far less computation time than conventional delay-and-sum methods. Two different reconstruction kernel functions have been derived and are compared for the processing of experimental data.

Contrast agent specific imaging modes for the ultrasonic assessment of parenchymal cerebral echo contrast enhancement

Previous work has demonstrated that cerebral echo contrast enhancement can be assessed by means of transcranial ultrasound using transient response second harmonic imaging (HI). The current study was designed to explore possible advantages of two new contrast agent specific imaging modes, contrast burst imaging (CBI) and time variance imaging (TVI), that are based on the detection of destruction or splitting of microbubbles caused by ultrasound in comparison with contrast harmonic imaging (CHI), which is a broadband phase-inversion-based implementation of HI. Nine healthy individuals with adequate acoustic temporal bone windows were included in the study. Contrast harmonic imaging, CBI, and TVI examinations were performed in an axial diencephalic plane of section after an intravenous bolus injection of 4 g galactose-based microbubble suspension in a concentration of 400 mg/mL. Using time-intensity curves, peak intensities and times-to peak-intensity (TPIs) were calculated off-line in anterior and posterior parts of the thalamus, in the region of the lentiform nucleus, and in the white matter. The potential of the different techniques to visualize cerebral contrast enhancement in different brain areas was compared. All techniques produced accurate cerebral contrast enhancement in the majority of investigated brain areas. Contrast harmonic imaging visualized signal increase in 28 of 36 regions of interest (ROIs). In comparison, TVI and CBI examinations were successful in 32 and 35 investigations, respectively. In CHI examinations, contrast enhancement was most difficult to visualize in posterior parts of the thalamus (6 of 9) and the lentiform nucleus (6 of 9). In TVI examinations, anterior parts of the thalamus showed signal increase in only 6 of 9 examinations. For all investigated imaging modes, PIs and TPIs in different ROIs did not differ significantly, except that TVI demonstrated significantly higher PIs in the lentiform nucleus as compared with the thalamus and the white matter (P < 0.05). The current study demonstrates for the first time that CBI and TVI represent new ultrasonic tools that allow noninvasive assessment of focal cerebral contrast enhancement and that CBI and TVI improve diagnostic sensitivity as compared with CHI.

New real-time strain imaging concepts using diagnostic ultrasound

Two real-time strain imaging concepts and systems are presented. Both systems are based on a conventional ultrasound scanner that is connected to a PC with an A/D converter card for real-time data acquisition of rf data. Differential strain between successively acquired rf frames are estimated using phase root seeking. The first concept uses a special real-time implementation of manual elastography. In the second concept, denoted 'vibrography', the static compression is replaced by low-frequency axial vibration of the probe, still operating in quasistatic acquisition mode. The properties of both concepts are discussed with regard to noise and motion artefacts, and it is shown, using simulations and phantom experiments, that both imaging concepts yield the same kind of strain images. Vibrography has the advantage that no manual compression has to be applied, total compression can be very low and some motion artefacts are better suppressed.

Axial strain imaging using a local estimation of the scaling factor from RF ultrasound signals

The main signal-processing techniques used in elastography compute strains as the displacement derivative. They perform well for very low deformations, but suffer rapidly from decorrelation noise. Aiming to increase the range of accurate strain measurements, we developed an adaptive method based on the estimation of strains as local scaling factors. Its adaptability makes this method appropriate for computing scaling factors resulting from larger strains or a wide spread of strain variations. First, segments corresponding to the same part of tissue are adaptively selected in the rest and stressed state echo signals. Then, local scaling factors are estimated by iteratively varying their values until reaching the zero of the phase of the complex cross-correlation function. Results from simulation and from experimental data are presented. They show how this adaptive method can track various local deformations and its accuracy for strain up to 7%.

Segmentation of 3D intravascular ultrasonic images based on a random field model

Segmentation of intravascular ultrasound images provides important information about the degree of vessel obstruction as well as about the shape and size of plaques. To address the problems of inter- and intra-observer variances associated with conventional manual tracing, a fully automated segmentation was developed. The algorithm is based on the optimisation of a maximum a posteriori estimator, implementing the Rayleigh distribution of speckle and a priori information about the contours. Within 3D image sets, additional information by the blood flow resulting in a decorrelation of the pixels within the luminal boundary is used to initialise the segmentation. To accelerate the estimation, dynamic programming was used. The segmentation algorithm was realised as a Windows 95 application on a Pentium II/233 MHz and delivered reliable and reproducible results independent of the catheter position and the total image brightness (except overflow). In contrast, contours drawn by two physicians for an evaluation of 29 clinical cases showed large intra- and inter-observer variances. In vivo images were acquired with a 20 MHz transducer array (EndoSonics InVision). Comparison with the contours drawn by the physicians and histology demonstrates the potential of the segmentation algorithm.

[Ultrasound elastography of the prostate. A new technique for tumor detection]

AIM

Prostate tumours are often of harder consistency than the surrounding tissue. During digital rectal examination, this fact can be used not only to detect hypertrophy but also localized hardenings. The examination by digital palpation is inaccurate and, even in combination with PSA-value and a transrectal ultrasonic examination, the result is often not reliable. Ultrasound elastography enables us to measure and visualize the elastic properties of a tissue region and is a useful supplement to the examination by digital palpation. Ultrasound elastography is able to measure and visualize the elastic properties of a tissue region, therefore it is a useful supplement to commonly used diagnostic procedures.

METHOD

We have developed a new system for elastographic prostate diagnosis which can be used during the transrectal ultrasonic examination. During the examination a sequence of ultrasonic images is acquired while the organ is slightly compressed by the ultrasound probe. Using numerical analysis of image pairs for the acquired sequence we calculate the tissue strain which represents the spatial elasticity distribution of a specific cross-section of the organ. This enables us to distinguish hard areas in the tissue. Image artifacts resulting from lateral motion components, i.e., orthogonal to the direction of the applied force, are compensated for by a special approach.

RESULTS

We present results obtained from a typical elastography phantom and also the first in vivo images from patients who were undergoing radical prostatectomy. Our images prove that ultrasound elastography has the potential to detect malignant tissue areas, which are inconspicuous in the B-mode image. Our findings are confirmed by the corresponding histological specimens.