Indentation Analysis of Biphasic Viscoelastic Hydrogels.

Mechanical properties of soft biological materials are dependent on the responses of the two phases of which they are comprised: the solid matrix and interstitial fluid. Indentation techniques are commonly used to measure properties of such materials, but comparisons between different experimental and analytical techniques can be difficult. Most models relating load and time during spherical indentation are based on Hertzian contact theory, but the exact limitation of this theory for soft materials are unclear. Here, we examine the response of gelatin hydrogels to shear and indentation loading to quantify combined effects of the solid and fluid phases. The instantaneous behavior of the hydrogels is different for each test geometry and loading rate, but the relaxed response, measured by the relaxed modulus, is the same for all tests, within 17%. Additionally, indentation depths from 15-25% of the radius of the spherical indenter are found to minimize error in the estimate of relaxed modulus.

pH-induced contrast in viscoelasticity imaging of biopolymers.

Understanding contrast mechanisms and identifying discriminating features is at the heart of diagnostic imaging development. This paper focuses on how pH influences the viscoelastic properties of biopolymers to better understand the effects of extracellular pH on breast tumour elasticity imaging. Extracellular pH is known to decrease as much as 1 pH unit in breast tumours, thus creating a dangerous environment that increases cellular mutation rates and therapeutic resistance. We used a gelatin hydrogel phantom to isolate the effects of pH on a polymer network with similarities to the extracellular matrix in breast stroma. Using compressive unconfined creep and stress relaxation measurements, we systematically measured the viscoelastic features sensitive to pH by way of time-domain models and complex modulus analysis. These results are used to determine the sensitivity of quasi-static ultrasonic elasticity imaging to pH. We found a strong elastic response of the polymer network to pH, such that the matrix stiffness decreases as pH was reduced; however, the viscous response of the medium to pH was negligible. While physiological features of breast stroma such as proteoglycans and vascular networks are not included in our hydrogel model, observations in this study provide insight into viscoelastic features specific to pH changes in the collagenous stromal network. These observations suggest that the large contrast common in breast tumours with desmoplasia may be reduced under acidic conditions, and that viscoelastic features are unlikely to improve discriminability.

Viscoelasticity imaging using ultrasound: parameters and error analysis.

Techniques are being developed to image viscoelastic features of soft tissues from time-varying strain. A compress-hold-release stress stimulus commonly used in creep-recovery measurements is applied to samples to form images of elastic strain and strain retardance times. While the intended application is diagnostic breast imaging, results in gelatin hydrogels are presented to demonstrate the techniques. The spatiotemporal behaviour of gelatin is described by linear viscoelastic theory formulated for polymeric solids. Measured creep responses of polymers are frequently modelled as sums of exponentials whose time constants describe the delay or retardation of the full strain response. We found the spectrum of retardation times tau to be continuous and bimodal, where the amplitude at each tau represents the relative number of molecular bonds with a given strength and conformation. Such spectra indicate that the molecular weight of the polymer fibres between bonding points is large. Imaging parameters are found by summarizing these complex spectral distributions at each location in the medium with a second-order Voigt rheological model. This simplification reduces the dimensionality of the data for selecting imaging parameters while preserving essential information on how the creeping deformation describes fluid flow and collagen matrix restructuring in the medium. The focus of this paper is on imaging parameter estimation from ultrasonic echo data, and how jitter from hand-held force applicators used for clinical applications propagate through the imaging chain to generate image noise.

Quantitative ultrasonic detection and classification of diffuse liver disease. Comparison with human observer performance.

A multiparameter ultrasonic tissue characterization system has been developed and tested on several types of diffuse liver disease. The four tissue characterization parameters used are based on the first and second order statistics of the B-scan image. Performance of the system was evaluated using receiver operating characteristic (ROC) analysis and was compared with the performance of experienced human observers viewing B-scan images. The machine-based multiparameter system achieved an area under the ROC curve (Az) of 0.88 for detection of chronic hepatitis in more than 100 proven cases of the disease. This was dramatically better than the performance of human observers (Az = .64, P less than .05) and compares favorably to the performance of other accepted diagnostic tests such as head CT and the PAP smear. For detection of Gaucher's disease, the Az for the system was .92, whereas for separating hepatitis from Gaucher's disease Az was .84. Human observers also did well at these tasks (P greater than .8) using organomegaly as their major criterion for diagnosing Gaucher's disease. For primary biliary cirrhosis the system Az was .80, for glycogen storage disease Az was .94. These results suggest that use of multiparameter tissue characterization can significantly increase the usefulness of ultrasound for evaluation of diffuse liver disease.

Renal ultrasound using parametric imaging techniques to detect changes in microstructure and function.

RATIONALE AND OBJECTIVES: Signal processing techniques have been used to generate parametric ultrasound images that describe properties of tissue microstructure. METHODS: Images of the average scatterer size (D) and integrated backscatter coefficient (IBC) for normal dog kidneys were examined. RESULTS: With parametric ultrasound the authors identified sources of cortical backscatter and observed microanatomical changes corresponding to ischemia. In particular, scatterer size images acquired in vitro and in vivo show it is possible to rapidly assess changes and differences in the average glomerular diameter and the average arteriolar cross-sectional diameter. CONCLUSIONS: A more direct interpretation of sonographic image data is possible with this new type of imaging. Parametric imaging may have a diagnostic role as a means to differentiate among conditions producing increased cortical echogenicity and to detect important structural indicators such as glomerular hypertrophy.

Quantitative ultrasonic detection of parenchymal structural change in diffuse renal disease.

OBJECTIVE: The authors determined whether quantitative ultrasound could be useful in the evaluation of diffuse renal disease. METHODS: Digitized radiofrequency ultrasound data were acquired from the kidneys of patients with biopsy-proven diffuse renal disease and transplant rejection (37 patients plus 18 normal volunteers). The results of the quantitative analysis were compared with histology results to determine if microscopic renal structure could be correlated with quantitative features such as scatterer size and scatterer spacing. The results also were analyzed using receiver operating characteristic analysis to determine if diffuse disease could be detected reliably using quantitative methods. RESULTS: The three most useful features in the native kidneys were mean scatterer spacing (MSS), sigma's, and average scatterer size (D). Using these features, it was possible to detect diffuse renal disease causing a decrease in renal function with an area under the ROC curve (Az) of 0.93. The feature D corresponded closely to histologically measured average glomerular diameters. For normals, D = 216 microns and glomerular diameter = 211 microns. No histologic correlate was found for scatterer spacing. In transplants, MSS and integrated backscatter were most useful for detecting rejection (Az = 0.87), and D in rejection was similar to the values for normal kidney and normally functioning transplants. CONCLUSIONS: The D value corresponds to glomerular diameter, and glomerular enlargement can be detected readily using quantitative ultrasound. Combinations of two to four quantitative features can detect diffuse renal disease and transplant rejection reliably.

B-mode registration: a new procedure for analyzing errors.

The theory of a new B-mode registration test procedure is presented along with experimental verification of the detailed theoretical predictions. The new procedure separately tests the B-scan articulated arm calibration and the equipment acoustic velocity calibration. Readily verifiable secondary standards are incorporated into the procedure for self-calibration purposes. The procedure provides a specific and reproducible analysis of angular misregistration errors while maintaining an acceptable level of sensitivity. No coupling gels or oils operator skill is required. A special test has been found which permits the service person to verify that the equipment is capable of attaining proper B-mode registration before beginning the recalibration procedure.

Acoustic backscattering from ultrasonically tissuelike media.

The acoustic backscattering coefficients of a tissue-mimicking (TM) material have been measured using a modified Sigelmann-Reid technique in order to assess the similarity of its acoustic properties to those of soft tissues. A simple reference scatter material was used to probe the strengths and the limitations of the measurement technique. Backscatter coefficients measured from the TM material exhibit a frequency dependence of the form fm in the range 1.0-8.0 MHz where 3.5 less than or equal to m less than or equal to 3.8. The backscatter coefficient is also found to be approximately proportional to the particle concentration in the TM material up to concentrations that yield an attenuation coefficient of 0.9 db/cm/MHz. Comparisons of backscatter coefficients measured from the TM material are made with published values for liver, myocardium, and blood.

Ultrasonically tissue-mimicking liver including the frequency dependence of backscatter.

Tissue-mimicking (TM) liver, previously reported by our laboratory, mimics the speed of sound, density, and attenuation coefficient (including frequency dependence) of liver; however, scatter properties are only qualitatively simulated. Two new versions of TM liver are reported here which not only mimic liver with respect to sound, density, and attenuation coefficient, but also with respect to backscatter coefficients, including the frequency dependence of the latter. Compositions, methods of production, and comparisons of ultrasonic properties with those found in the literature for human liver are presented.

Contrast-detail analysis of image degradation due to lossy compression.

A contrast-detail (CD) experiment was performed to study the effect of lossy compression on computed radiographic (CR) images. Digital CR images of a phantom were compressed by quantizing the full-frame discrete cosine transform and Huffman encoding the result. Since low-contrast detectability is directly linked to an important radiological task, namely, the detection of noncalcified pulmonary nodules in adult chest radiographs, the goal of the study was to quantify any loss in low-contrast detectability due to compression. Compression ratios varied significantly among compressed images, despite the use of fixed compression parameters; detectability could be specified by a single parameter of a CD curve; there was no significant reduction in detectability for an average compression ratio of 11:1; and, there was a statistically significant degradation in detectability for an average compression ratio of 125:1.

Ultrasound contrast-detail analysis: a comparison of low-contrast detectability among scanhead designs.

Contrast-detail (CD) analysis was used to compare the low-contrast detection capabilities of expert observers using different array-type scanhead technologies. Five expert observers viewed five different contrast targets to obtain CD curves for each scanhead. Differences in CD curves are interpreted in terms of the image contrast, resolution, and noise. It was found that differences in low-contrast detectability were due to differences in beam properties. Clinical images obtained during patient examinations are used to show how some clinically relevant tasks are distributed in their contrast and size.

Ultrasound contrast-detail analysis: a preliminary study in human observer performance.

Results of a human observer performance study using a new ultrasound contrast-detail (CD) phantom are reported. The new phantom allows estimates of the fundamental statistical uncertainties involved in CD analysis. Results of these experiments show that observers generally considered to be "trained" require experience at the specific task of low-contrast target detection under signal known exactly (SKE) conditions, and that observers require an orientation period at the beginning of each observation session to obtain stable responses. The results obtained in this study can be used to estimate the number of independent images, observers, and repeat observation sessions required for a desired uncertainty in CD curves, and show that 10% fractional standard deviations are obtainable with modest observer effort.

Comparison of the low-contrast detectability of a screen-film system and third generation computed radiography.

Contrast-detail (CD) analysis was used to compare the low-contrast detectability of computed radiography (CR) and screen-film (SF) as applied to the task of adult chest radiography. A phantom was constructed and imaged using the same exposure factors throughout all experiments. Within-observer variance, between-observer variance, and image sample variance were calculated and used to estimate the standard error for each experiment. The results of these CD experiments agreed with the predictions of the Rose model. Observers performed equally well for low-contrast target detection using CR and SF.

The effects of lossy compression on the detection of subtle pulmonary nodules.

We examined the ability of radiologists to detect pulmonary nodules in computed radiographic (CR) chest images subjected to lossy image compression. Low-contrast 1-cm diameter targets simulating noncalcified pulmonary nodules were introduced into clinical images and presented to ten radiologists in a series of two-alternative forced-choice (2AFC) observer experiments. The percentages of correct observer responses obtained while viewing noncompressed images (1:1) were compared with those obtained for the same images compressed 7:1, 16:1, 44:1, and 127:1. The images were compressed using a standard full-frame discrete cosine transform (DCT) technique. The degree of compression was determined by quantizing Fourier components in various frequency channels and then Huffman encoding the result. The data show a measurable decline in performance for each compression ratio. Through signal-to-noise ratio (SNR) analysis, we found that the reduction in performance was due primarily to the compression algorithm that increased image noise in the frequency channels of the signals to be detected.

Comparison of a cathode-ray-tube and film for display of computed radiographic images.

The goal of the study was to compare a cathode-ray-tube (CRT) digital display with film by using task-dependent image quality assessment methods. Contrast-detail analysis was utilized. Human observers performed a simple detection task, specifically, detecting a pillbox target in a uniform Poisson field, using either film or a digital display that employed a CRT monitor. Observers performed equally well on both film and CRT when the window settings of the digital display were established subjectively by a radiologist. Changing the window settings of the digital display to match the average background luminance of a film-illuminator combination decreased the luminance contrast of the targets and observer performance was reduced, though these effects were probably not linked. The "gold standard" film had lower luminance contrast than the CRT displayed images, yet observer performance was never lower for film than for the CRT. Therefore we concluded that luminance contrast was not a limiting factor for observer performance in this study. The CRT monitor changed fairly rapidly after it was calibrated. During a period of six months the gamma of the display increased from 1.82 to 2.42 and the maximum luminance decreased from 319 to 228 cd/m2. Low luminance output demonstrated a larger percentage decrease (approximately equal to 85%) than high luminance output (approximately equal to 29%) over the same time period. These observations suggest that standard window settings should be reviewed from time to time to ensure that the display is used optimally. No special look-up table setup such as perceptual linearization was used.

Errors in biased estimators for parametric ultrasonic imaging.

Maximum likelihood (ML) methods are widely used in acoustic parameter estimation. Although ML methods are often unbiased, the variance is unacceptably large for many applications, including medical imaging. For such cases, Bayesian estimators can reduce variance and preserve contrast at the cost of an increased bias. Consequently, including prior knowledge about object and noise properties in the estimator can improve low-contrast target detectability of parametric ultrasound images by improving the precision of the estimates. In this paper, errors introduced by biased estimators are analyzed and approximate closed-form expressions are developed. The task-specific nature of the estimator performance is demonstrated through analysis, simulation, and experimentation. A strategy for selecting object priors is proposed. Acoustic scattering from kidney tissue is the emphasis of this paper, although the results are more generally applicable.

Image staining and differential diagnosis of ultrasound scans based on the Mahalanobis distance.

The covariance matrices associated with each state of health or disease from a previous study are used as the basis of an image staining display technique for aid in quantitative differential diagnosis. A state of health or disease is chosen by the clinician: this selects the covariance matrix from the data base. A region of interest (ROI) is then scrolled through an abdominal B-scan. For each position of the ROI a point in the four-dimensional feature space is calculated. A natural measure of the distance of this point from the center of mass (multivariate mean) of the disease class is calculated in terms of the covariance matrix of this class; this measure is the Mahalanobis distance. The confidence level for acceptance or rejection of the hypothesized disease class is obtained from the probability distribution of this distance, the T(2) probability law. This confidence level is color coded and used as a color stain that overlays the original scan at that position. The variability of the calculated features is studied as a function of ROI size, or the spatial resolution of the color coded image, and it is found that for an ROI in the neighborhood of 4 cm(2) most of the variability due to the finite number of independent samples (speckles) is averaged out, leaving the "noise floor" associated with inter- and intra-patient variability. ROIs on the order of 1 cm(2) may result with technical advances in B-scan resolution. A small number of points on organ boundaries are entered by the user, to fit with arcs of ellipses to be used to switch between organ (liver and kidney) data bases as the ROI encounters the boundary. By selecting in turn various state-of-health or state-of-disease databases, such images of confidence levels may be used for quantitative differential diagnosis. The method is not limited to ultrasound, being applicable in principle to features obtained from any modality or multimodality combination.

Characterising the microstructure of random media using ultrasound.

This paper describes a technique of forming images of the average size and scattering strength of scatterers in a random medium using ultrasound. Quantitative ultrasound images provide a more direct interpretation of the underlying structure of the medium, e.g. size, shape, number and elastic properties of scatterers, and increased detectability for regions of varying structure. A signal-to-noise analysis was used to show quantitatively how properties of the imaging system influence low-contrast detectability in quantitative ultrasound images. In one experiment, signal-to-noise measurements using phantoms were compared with B-mode imaging for several transducer bandwidths to observe variations in image contrast and speckle noise. The findings are being used to optimise the design of quantitative imaging systems for specific diagnostic tasks.

Elastostatics of a spherical inclusion in homogeneous biological media.

A three-dimensional spherical inclusion model that approximates a lesion bonded to a tissue matrix is proposed for biomedical elastography. Analytical formulae describing spatial strain and stress distributions generated in infinite media by uniform loading are given under a linear, homogeneous, isotropic elasticity assumption. Strain and stress distributions are also calculated using finite-element analysis (FEA) for a variety of cases to determine the effects of shear modulus distribution, external loading conditions (uniform stress versus uniform displacement), compressor size and matrix dimensions on the elastostatics of the tissue. Analytical strain and stress predictions are shown to agree with the FEA results to within 10% accuracy provided that the matrix dimensions are at least ten times that of the inclusion. Also for these cases, uniform-stress boundary conditions can be equivalently represented by uniform displacement of the boundary. Spherical inclusions exhibit a lower efficiency for transferring elastic shear modulus contrast into strain contrast than cylindrical or planar inclusions. Additional compression will increase the strain contrast. However, large compressions also lead to increases in ultrasonic signal decorrelation and strain and stress concentrations in the homogeneous matrix around the inclusion. Although strain concentrations may help describe the boundaries of the inclusion more clearly, they also increase the risk of damaging the tissue. Understanding the strain and stress distributions in a biological tissue containing a lesion is necessary for optimizing the experimental configurations and consequently improving the diagnostic values of elasticity imaging.

New contrast-detail phantoms for improved precision in lesion detection measurements.

The previous design of our ultrasound contrast-detail (C-D) phantom is limited in its ability to evaluate the quality of diagnostic ultrasound imaging systems. There is uncertainty in the contrast-detail measurements due to the single available viewing angle for each target and the resulting single realization of the ultrasound speckle pattern. Two new contrast-detail phantom designs are described which enable many independent realizations of the speckle noise for observer C-D experiments of improved precision. In the first design, a single tissue-mimicking cone is located on the axis of a tissue-mimicking cylinder. Cross-sectional images of the cone which simulate focal lesions can be obtained from any orientation by rotating the cylinder under the transducer. In the second new design, a tissue-mimicking cone is positioned in a tissue-mimicking slurry of agar/graphite spheres. Gentle stirring of the slurry and rotation of the cone produce many independent realizations of the speckle. In both new phantoms, the lesion contrast can be specified and is frequency/transducer independent.