In medical diagnosis, use of elastography is becoming increasingly more useful. However, treatments usually
assume a planar compression applied to tissue surfaces and measure the deformation. The stress distribution
is relatively uniform close to the surface when using a large, flat compressor but it diverges gradually along
tissue depth. Generally in prostate elastography, the transrectal probes used for scanning and compression are
cylindrical side-fire or rounded end-fire probes, and the force is applied through the rectal wall. These make it
very difficult to detect cancer in prostate, since the rounded contact surfaces exaggerate the non-uniformity of
the applied stress, especially for the distal, anterior prostate.
We have developed a preliminary 2D Finite Element Model (FEM) to simulate prostate deformation in
elastography. The model includes a homogeneous prostate with a stiffer tumor in the proximal, posterior region
of the gland. A force is applied to the rectal wall to deform the prostate, strain and stress distributions can
be computed from the resultant displacements. Then, we assume the displacements as boundary condition and
reconstruct the modulus distribution (inverse problem) using linear perturbation method.
FEM simulation shows that strain and strain contrast (of the lesion) decrease very rapidly with increasing
depth and lateral distance. Therefore, lesions would not be clearly visible if located far away from the probe.
However, the reconstructed modulus image can better depict relatively stiff lesion wherever the lesion is located.
We report preliminary results from our investigation of in vivo prostate elastography. Fewer than 50% of all prostate cancers are typically visible in current clinical imaging modalities. Elastography displays a map of strain that results when tissue is externally compressed. Thus, elastography is ideal for imaging prostate cancers because they are generally stiffer than the surrounding tissue and stiffer regions usually exhibit lower strain in elastograms. In our study, digital radio-frequency (RF) ultrasound echo data were acquired from prostate-cancer patients undergoing brachytherapy. Seed placement is guided by a transrectal ultrasound (TRUS) probe, which is held in a mechanical fixture. The probe can be moved in XYZ directions and tilted. The probe face, in contact with the rectal wall, is used to apply a compression force to the immediately adjacent prostate. We also used a water-filled (acoustic) coupling balloon to compress the prostate by increasing the water volume inside the balloon. In each scan plane (transverse), we acquired RF data from successive scans at the scanner frame rate as the deformation force on the rectal wall was continuously increased. We computed strain using 1D RF cross-correlation analysis. The compression method based on fixture displacement produced low-noise elastograms that beautifully displayed the prostate architecture and emphasized stiff areas. Balloon-based compression also produced low-noise elastograms. Initial results demonstrate that elastography may be useful in the detection and evaluation of prostate cancers, occult in conventional imaging modalities.
We describe a novel and robust strain estimation method, which is capable of fast, accurate strain estimation even in the presence of large signal decorrelation. Global temporal stretching of post-compression signals compensating for the applied strain significantly improves the "quality" of strain estimates. In a natural extension of this approach (adaptive stretching), a search is performed at each data window for the stretch factor that maximizes the correlation between the pre- and post-compression echo signal segments. Adaptive stretching performs well under harsh signal environments (because the correlation is maximized at each location); however it is computation intensive because many iterations may be required at each location. In contrast, global stretching is a fast algorithm, but performs well only in areas where local strains are close to the applied strain. The proposed method strikes a balance between the speed of global stretching and the performance of adaptive stretching. In this method, global stretching is performed with a range of different stretch factors and strain maps are computed for each stretch factor. The correlation between the pre- and post-compression echo segments is the maximum when the stretch factor corresponds to the local strain. Thus, the area in each computed strain image with strain values closely corresponding to the uniform stretch factor will contain "good quality" strain estimates. (Naturally, this area is different in each image.) To produce a single elastogram at the end, these strain maps are combined as follows. Correlation values quantify the "quality" of strain estimates; thus, at each location we identify the strain map with the maximum correlation, and the strain value in that strain map at that location is chosen for the combined map. Results from data generated by finite-element simulation and phantom experiments show that the variable stretching strain estimator is fast and is significantly less susceptible to signal degradation than conventional strain estimators.
We describe a novel strain estimation method, which is capable of accurate strain estimation in the presence of large or irregular tissue motion. In conventional elastography, tissue strains induced by external compression applied to the tissue surface, are estimated by cross-correlation analysis of echo signals obtained before and after compression. Large and irregular tissue motions significantly change echo-signal shapes, causing major echo-signal decorrelations. In the presence of significant signal decorrelations, the estimated displacements may have many discontinuities because of false peak errors (primary correlation peak smaller than a secondary peak). However, because tissue is virtually incompressible, true elastographic displacements have spatial continuity in both axial and lateral directions, and the true correlation peaks, albeit diminished, are still present and detectable in the presence of spurious peaks. Our approach treats the total ensemble of correlation functions vs. depth and use information from neighboring areas to remove ambiguity that results from false peak errors. Preliminary results using finite-element simulation show that the correlation-tracking strain estimator (CTSE) can produce excellent strain estimates in harsh environments.
Brachytherapy using small implanted radioactive seeds is becoming an increasingly popular method for treating prostate cancer. Seeds are inserted into the prostate transperineally using ultrasound guidance. Dosimetry software determines the optimal placement of seeds for achieving the prescribed dose based on ultrasonic determination of the gland boundaries. However, because of prostate movement after planning images are acquired and during the implantation procedure, seeds commonly are not placed in the desired locations and the delivered dose may differ from the prescribed dose. Current methods of ultrasonic imaging do not adequately display implanted seeds for the purpose of correcting the delivered dose. We are investigating new methods of ultrasonic imaging that overcome limitations of conventional ultrasound. These methods include resonance, modified elastographic, and signature techniques. Each method shows promise for enhancing the visibility of seeds in ultrasound images. Combining the information provided by each method may reduce ambiguities in determining where seeds are present or absent. If successful, these novel imaging methods will enable correction of seed-misplacement errors during the implantation procedure, and hence will improve the therapeutic radiation dose delivered to target tissues.
We have developed a family of objective features in order to provide non-invasive, reliable means of distinguishing benign from malignant breast lesions. These include acoustic features (echogenicity, heterogeneity, shadowing) and morphometric features (area, aspect ratio, border irregularity, margin definition). These quantitative descriptors are designed to be independent of instrument properties and physician expertise. Our analysis included manual tracing of lesion boundaries and adjacent areas on grayscale images generated from RF data. To derive quantitative acoustic features, we computed spectral parameter maps of radio-frequency (RF) echo signals (calibrated with system transfer function and corrected for diffraction) within these areas. We quantified morphometric features by geometric and fractal analysis of traced lesion boundaries. Although no single parameter can reliably discriminate cancerous from non-cancerous breast lesions, multifeature analysis provides excellent discrimination of cancerous and non-cancerous lesions. RF echo-signal data used in this study were acquired during routine ultrasonic examinations of biopsy-scheduled patients at three clinical sites. Our data analysis for 130 patients produced an ROC-curve area of 0.9164 +/- 0.0346. Among the quantitative descriptors, lesion heterogeneity, aspect ratio, and a border irregularity descriptor were the most useful; some morphometric features (such as the border irregularity descriptor) were particularly effective in lesion classification.
In conventional elastography, strains are estimated by computing gradient of estimated displacement. However, gradient-based algorithms are susceptible to noise. We have developed two new strain estimators to overcome the common limitations of elastography. The first estimator is based on a frequency-domain formulation; it estimates local strain by maximizing the correlation between the spectra of pre- and post-compression echo signals by iteratively frequency- scaling the latter. We discuss a variation of this algorithm that may be computationally more efficient. The second estimator is based on the observation that an extremely stiff region will undergo virtually no strain when compressed, and will exhibit quasi-rigid body motion. As a result, an area with high similarity between the pre- and post-compression signals indicates low strain, and an area with low similarity indicates large strain. We use normalized 2D correlation function to estimate this similarity. This method offers significant advantages for detecting rigid tissues in the presence of large, irregular, non-axial motion. Both the estimators exhibited promising results in simulation and experiments.
We are developing quantitative descriptors of breast lesions in order to provide reliable, operator-independent means of non-invasive breast cancer identification. These quantitative descriptors include lesion internal features assessed using spectrum analysis of ultrasonic radio-frequency (RF) echo signals and morphometric features related to lesion shape. Internal features include quantitative measures of 'echogenicity,' 'heterogeneity,' and 'shadowing;' these were computed by generating spectral-parameter images of the lesion and surrounding tissue. Spectral-parameter values were generated at each pixel in the parameter image using a sliding-window Fourier analysis. Lesions were traced on B-mode images and traces were used in conjunction with spectral parameter values to compute echogenicity, heterogeneity, and shadowing. Initial results show that no single parameter may be sufficiently precise in identifying cancerous breast lesions; the results also show that the use of multiple features can substantially improve discrimination. This paper describes the background, research objective, and methodology. Clinical examples are included to illustrate the practical application of our methodology.