An evaluation of our experience in position verification of catheters used for interstitial high-dose-rate brachytherapy of solitary bladder tumors.

PURPOSE: The goal of this study was to verify the position of catheters used over 4 days for brachytherapy of solitary bladder tumors. METHODS AND MATERIALS: The study covered three phases. Shifting of catheters was studied using daily position verification CT scans of 20 patients. The possibility to omit the CT scan on Day 2 by adding a loading margin of 4 mm on each side was studied using data of 5 patients. Whether the Day 4 verification CT scan could be omitted if this margin was used, was studied for another group of 10 patients, comparing the Day 3 treatment plan to the Day 4 CT scan. RESULTS: An average catheter shift on Days 2, 3, and 4 of, respectively, -0.3 mm (-8 to 10 mm), -0.5 mm (-14 to 10 mm), and -0.1 mm (-16 to 28 mm) was found over the measurements at both sites of the catheter. Including only shifts causing underdosing of the clinical target volume (CTV), the average shift on Days 2, 3, and 4 was, respectively, -3.6 mm (-1 to -8 mm), -5.4 mm (-1 to -14 mm), and -5.3 mm (-1 to -16 mm). After adding a loading margin, the CTV was covered on Day 2; however, the margin was not sufficient for Days 3 and 4. On Day 4, in 2/10 patients, the CTV was not completely covered. In 5/10 patients, an increased 200% isodose volume was found. CONCLUSIONS: Position verification is necessary in bladder brachytherapy. If a 4-mm margin on each side of the loading pattern was added, position verification on Day 2 could be omitted. The verification CT scan of Days 3 and 4 is still necessary.

Irradiation and responsiveness to pain stimuli in rats.

This study evaluates whether irradiation inhibits responses to pain in an animal model. We found that irradiation with doses of 10 Gy, 15 Gy and 17.5 Gy of the lumbar enlargement of the spinal cord inhibits the behavioural responses to the stimulus of the hot-plate. These doses were otherwise without effects. This data is discussed in view of the effects of irradiation of living cells, and we propose that a modification of pain signal processing is accomplished. Similar considerations apply to the human condition.

Correlations between acoustic and texture parameters from RF and B-mode liver echograms.

Radio frequency (RF) echograms were acquired from human subjects without liver pathology (n = 126), who were in the range of 20 to 84 years of age. After appropriate correction for the equipment settings and performance characteristics, acoustospectrographic parameters were estimated. The data were corrected for the frequency-dependent attenuation and then software demodulated. Image texture parameters were calculated based on the assumption of a particular tissue model, and purely statistical parameters were also estimated. The data thus obtained from each subject were corrected for the age trend that was assessed in an earlier publication by the authors. A total of 18 parameters was considered and the correlations were analyzed. It is concluded that nine parameters could be identified which did not strongly correlate with each other. This conclusion is supported by analysis of correlation data from patients. In a companion paper, a discriminant analysis is reported, based on the selected parameters (nine) and applied to the differentiation between normals and various classes of diffuse liver disease.

Detection of diffuse liver disease by quantitative echography: dependence on a Priori choice of parameters.

Quantitative acoustic parameters and image texture parameters were used in a linear discriminant analysis. This analysis was applied to detect retrospectively the classes of diffuse liver disease against a population of normal livers. Three different sets of parameters were employed. The first set was selected by the authors, and the other two were taken from the literature. The area under the Receiver Operating Characteristic (ROC) (or percentage correct classification) obtained with the first set ranged from 88% to 97%, depending on the disease class. It is concluded that the first-order statistical parameters of the image texture (diffuse scattering model) together with the slope of the attenuation coefficient are the most important parameters. As an alternative to the texture parameters, the backscattering parameters (second set of parameters) also yielded a comparably high score. The texture analysis involving structural scattering (third set of parameters) produced a lower percentage of correct classification. The overall conclusion is that the methods devised might be used for prospective diagnosis.

Detection and differentiation of diffuse liver disease by quantitative echography. A retrospective assessment.

RATIONALE AND OBJECTIVES: The detectability of diffuse liver diseases by quantitative echography was retrospectively investigated using scans of patients with known pathologic findings (n = 103) and of normal subjects (n = 129). The authors determined the best set of quantitative parameters for this task. METHODS: Quantitative echography was comprised of acoustospectrographic parameters (frequency dependence of attenuation and backscattering) and image texture parameters. The disease processes studied included: acute hepatitis, hepatitis/cirrhosis, alcoholic hepatitis/cirrhosis, primary biliary cirrhosis, and steatosis. RESULTS: Correct differentiation of these diseases ranged from 88% to 97%. Correlations between histologic grading and echographic parameters were poor. With only one exception, the differentiation between any two of the diseases could be made in 60% to 99% of cases. Different parameters better differentiated abnormal from normal scans than among diseases. CONCLUSIONS: The detection of diffuse liver diseases can be based on echographic parameters, related to a diffuse scattering model, whereas the differentiation among diseases needs additional parameters derived from a structural scattering model. Further studies are indicated to assess the prospective potential of the devised methods.

Ultrasound attenuation and texture analysis of diffuse liver disease: methods and preliminary results.

A study was performed to find and test quantitative methods of analysing echographic signals for the differentiation of diffuse liver diseases. An on-line data acquisition system was used to acquire radiofrequency (RF) echo signals from volunteers and patients. Several methods to estimate the frequency-dependent attenuation coefficient were evaluated, in which a correction for the frequency and depth-dependent diffraction and focusing effects caused by the sound beam was applied. Using the estimated value of the attenuation coefficient the RF signals themselves were corrected to remove the depth dependencies caused by the sound beam and by the frequency-dependent attenuation. After this preprocessing the envelope of the corrected RF signals was calculated and B-mode images were reconstructed. The texture was analysed in the axial direction by first- and second-order statistical methods. The accuracy and precision of the attenuation methods were assessed by using computer simulated RF signals and RF data obtained from a tissue-mimicking phantom. The phantom measurements were also used to test the performance of the methods to correct for the depth dependencies. The echograms of 163 persons, both volunteers and patients suffering from a diffuse liver disease (cirrhosis, hepatitis, haemochromatosis), were recorded. The mutual correlations between the estimated parameters were used to preselect parameters contributing independent information, and which can subsequently be used in a discriminant analysis to differentiate between the various diseased conditions.

Variability of quantitative echographic parameters of the liver: intra- and interindividual spread, temporal- and age-related effects.

The values of acoustic and image texture parameters were estimated from liver scans of healthy subjects. The values were obtained after appropriate preprocessing of the radio frequency echograms by an on-line computerized system. The preprocessing comprised a correction for the Time Gain Compensation (TGC), the beam diffraction and the frequency dependent attenuation in the Region of Interest (ROI). The intra- and interindividual variability of the parameter values appeared to be of the same order of magnitude, but significantly larger than the variability assessed by measurements of a homogeneously scattering tissue mimicking phantom. Significant temporal effects were found for all the parameters, which consistently occurred during the morning. These results are discussed in relation to the circadian rhythm of the glycogen content and of the hepatic circulation. All the parameters appeared to be significantly correlated to age. The slope of the regression ranged from 3.6% per decade (attenuation coefficient) to 7.6% per decade (mean echo-level). A tentative explanation to these results is presented: the increased stiffness of hepatic vasculature with age.

Ultrasonic differentiation of intraocular melanomas: parameters and estimation methods.

In this study, the estimation of ultrasound parameters is evaluated for in vivo differentiation of intraocular melanomas. For this purpose, both tissue and image parameters of the ultrasound signal are considered. These parameters comprised, respectively, the frequency dependent attenuation and backscattering coefficient of the melanoma tissue, and the first and second-order statistics of the amplitude-modulated and phase-derivative images of the melanomas. A diffraction correction procedure has been applied prior to the estimation of the parameters to correct the ultrasound signals for the echographic equipment used and for the various distances of the region-of-interest to the transducer. In addition, a pre-processing to select a homogeneous region from the tumours was implemented to obtain consistent estimates of the ultrasound parameters, because the accuracy and the precision of the parameters would be greatly reduced by the inhomogeneity of the melanoma tissue. The estimation methods are evaluated by means of the accuracy and precision of the parameters estimated from simulated ultrasound data and data obtained from a tissue-mimicking phantom. The mutual correlations of the parameters are discussed for the ultrasound data obtained from the melanomas. This study enabled a preselection of the independent ultrasound parameters that could be used in a discriminant analysis to perform a differentiation of intraocular melanomas. The sensitivity and specificity of differentiating spindle cell type from mixed-epitheloid meleanomas were 92 and 89 percent, respectively.

Texture in tissue echograms. Speckle or information?

Models of biological tissues are described in terms of acoustic parameters and of structure. Beam formation is discussed for continuous wave and pulsed modes of transducer operation and the concept of the point spread function (PSF) is introduced. The PSF is equivalent to the resolution cell, or the sampling volume, of echographic equipment. The generation of echograms from parenchymal tissues is described in terms of speckle formation due to interference at reception on the transducer. The speckle dimensions are quantitatively compared to the sampling volume of the employed transducer. It is shown that for fully developed speckle the tissue characteristics are exclusively reflected in the mean echolevel and not in the speckle size. The speckle size is, however, greatly dependent on the bandwidth, the frequency, and the geometry of the employed transducer. The attenuation by the insonated tissue yields a depth-dependent increase of mainly the lateral speckle size, in addition to the depth dependence caused by the beam formation. If the number density of scattering sites within the tissue is relatively low, the speckle characteristics are dependent on this density and, hence, tissue characterization is feasible if these characteristics are analyzed by statistical methods. These methods are gray level histogram analysis and the estimation of the autocorrelation function, ie, first and second order statistics, respectively. Structural order in tissues can be quantified by autocorrelation analysis and clinical studies on diffuse liver diseases support this conclusion. The effects of pre- and postprocessing on the detectability of focal lesions are outlined. The impact of multifocus systems and of the acquisition of radio frequency echograms on further developments of clinical echography is discussed.

Phase-derivative imaging. III: Theoretical derivation of first and second order statistics.

The echo signal obtained from a homogenous and isotropically scattering medium can be described as a Poisson time series which is convolved with the transmission pulse of the transducer. The probability density function (pdf) of this signal approximates to a Gaussian pdf for narrowband pulse waveform. Methods to derive the phase-derivative (PD) signal from the complex envelope and the preenvelope of the echo signal are described. The first order pdf of the PD asymptotically becomes a Gaussian pdf by smoothing. Since the rectified PD is employed to obtain 2-dimensional grey scale images, the first order pdf as well as the signal-to-noise ratio (SNR) of this signal are also derived. The rectified PD is further smoothed by a cosine time window prior to the imaging. The SNR and the autocorrelation function (in the axial direction) of this latter signal can be derived under the assumption of a Gaussian spectrum of the transmission pulse. These first and second order characteristics of the PD images are calculated for the conditions employed in simulations and experiments reported previously and are quantitatively compared to the values obtained from these.

Phase-derivative imaging. II: Effects of beam diffraction and scatterer density.

The potential of using the phase derivative (PD) of the radio-frequency echograms for producing 2-dimensional grey scale images was further investigated. The PD images were produced by five different algorithms, which according to the results described in the companion paper yield PD images dominated by the amplitude (envelope, ENV), mixed PD-AM images and pure PD images. These images are termed according to their algorithms: ZCS, zero crossing counter with squelch; ASS, analytic signal with squelch; ASW, analytic signal with Wiener kernel; UNP, unwrapped phase; and SAS, smoothed analytic signal. The rf data were obtained from simulations and from experiments with a tissue mimicking phantom. PD images were analysed by calculation of the first and second order grey level statistics: mean level, signal-to-noise ratio (SNR) and the full width at half maximum (FWHM) of the autocovariance functions (ACVF). These parameters were systematically investigated for a range of depths with respect to the transducer and a range of scatterer densities of the insonated medium. The UNP and SAS images do not suffer much from the diffraction effect but do not display much information about the scatterer density either. The ASW and ASS images qualitatively display beam diffraction effects similar to those of the AM images, with the exception of the mean value which is at the minimum in the focus, where the AM yields a sharp maximum at that depth. The mean and the SNR of the ASW and ASS images increase with increasing scatterer density and saturate at a density of 5000 cm-3. The mean value of the envelope, however, displays a square root dependency over the whole range. The axial and lateral FWHM of the ACVF of the UNP and SAS methods are not significantly dependent on the scatterer density and decrease with increasing density for the ASW and ASS images, as was observed in the envelope images. It may be concluded that ASW and ASS methods produce PD grey scale images which are equally well suited for the diagnosis of diffuse diseases of parenchymal tissues as conventional AM images. The smaller "speckle" size of the ASW images might be advantageous, for the detection of focal lesions, but the lesion contrast is found to be much lower than for the ENV.

Gray level transforms and lesion detectability in echographic images.

In search of the optimal display of echographic information for the detection of focal lesions, a systematic study was performed considering a wide range of gray level transforms (i.e., lookup tables). This range comprised power functions of the echo envelope signal (1/8 less than or equal to n less than or equal to 8), power functions of the logarithmic transform and a sigmoid function. The implications of the transforms on the first order statistics (histogram, "point signal-to-noise ratio" SNRp) and on the second order statistics (autocorrelation function) could be derived both analytically, and from the analysis of simulated and experimentally obtained echograms of homogeneously scattering tissue models. These results were employed to estimate the lesion signal-to-noise ratio SNRl, which specifies the detectability of a lesion by an ideal observer. It was found, both theoretically and practically, that the intensity display corresponds to the optimal transform (i.e., n = 2) for a low contrast lesion. When the data were first logarithmically compressed, the lesion SNR appeared to increase with increasing power (1/8 less than or equal to n less than or equal to 8). A logarithmic transform followed by a sigmoid compression did not produce much improvement. These effects of gray level transforms on the SNRl were shown to be relatively small, with the exception of powers n greater than 2 when applied to linear (i.e. amplitude) data. In the case of high lesion contrast, the sequence of log compression, followed by a square law produced the optimum SNRl. This sequence is equivalent to the processing within echographic equipment, where the TV monitor has a gamma of the order of 2.

Phase-derivative imaging. I: Methods and stabilization analysis.

The potential of using the phase derivative (PD) for echographic imaging was investigated. The PD data were calculated by four methods: zero crossing (ZCS) with squelch addition, analytic signal either with squelch addition (ASS) or with employment of a Wiener kernel (ASW), and unwrapped phase (UWP). The large peaks which occur in an unprocessed PD signal were "stabilized" by some kind of smoothing algorithm. The effects of the amplitude of the squelch signal and of the degree of smoothing were systematically investigated for experimental and simulated 1-D and 2-D rf echograms. The optimal pictures obtained for all four PD estimation methods were compared to the amplitude modulated (AM) image obtained from the same rf data. It is concluded that three different PD images can be derived: AM dominated (ZCS, ASS), mixed AM-PD (ASW) and pure PD (UWP) images. Some preliminary conclusions regarding the potential of PD imaging for medical diagnostics were drawn. These conclusions were based on quantitative 1st order statistics and on a qualitative assessment of 2nd order statistics of the PD image texture.

Texture of B-mode echograms: 3-D simulations and experiments of the effects of diffraction and scatterer density.

B-mode echograms were simulated by employing the impulse response method in transmission and reception using a discrete scatterer tissue model, with and without attenuation. The analytic signal approach was used for demodulation of the RF A-mode lines. The simulations were performed in 3-D space and compared to B-mode echograms obtained from experiments with scattering tissue phantoms. The average echo amplitude appeared to increase towards the focus and to decrease beyond it. In the focal zone, the average amplitude increased proportionally to the square root of the scatterer density. The signal to noise ratio (SNR) was found to be independent of depth, i.e., 1.91 as predicted for a Rayleigh distribution of gray levels, although a minimum was found in the focal zone at relatively low scatterer densities. The SNR continuously increased with increasing scatterer density and reached the limit of 1.91 at relatively high densities (greater than 10(4) cm-3). The lateral full width at half maximum (FWHM) of the two dimensional autocovariance function of the speckle increased continuously from the transducer face to far beyond the focus and decreased thereafter due to the diffraction effect. The lateral FWHM decreased proportionally to the logarithm of the scatterer density at low densities and reached a limit at high densities. Introduction of attenuation in the simulated tissue resulted in a much more pronounced depth dependence of the texture. The axial FWHM was independent of the distance to the transducer to a first approximation and decreased slightly with increasing scatterer density until a limit was reached at densities larger than 10(3) cm-3. This limit was in agreement with theory. The experiments confirmed the simulations and it can be concluded that the presented results are of great importance to the understanding of B-mode echograms and to the potential use of the analysis of B-mode texture for tissue characterization.