Tissue Regions Of Interest Research Articles

A Modular Software System to Assist Interpretation of Medical Images—Application to Vascular Ultrasound Images

Improvements in medical imaging technology have greatly contributed to early disease detection and diagnosis. However, the accuracy of an examination depends on both the quality of the images and the ability of the physician to interpret those images. Use of output from computerized analysis of an image may facilitate the diagnostic tasks and, potentially improve the overall interpretation of images and the subsequent patient care. In this paper, Analysis, a modular software system designed to assist interpretation of medical images, is described in detail. Analysis allows texture and motion estimation of selected regions of interest (ROIs). Texture features can be estimated using first-order statistics, second-order statistics, Laws' texture energy, neighborhood gray-tone difference matrix, gray level difference statistics, and the fractal dimension. Motion can be estimated from temporal image sequences using block matching or optical flow. Image preprocessing, manual and automatic definition of ROIs, and dimensionality reduction and clustering using fuzzy c-means, are also possible within Analysis. An important feature of Analysis is the possibility for online telecollaboration between health care professionals under a secure framework. To demonstrate the applicability and usefulness of the system in clinical practice, Analysis was applied to B-mode ultrasound images of the carotid artery. Diagnostic tasks included automatic segmentation of the arterial wall in transverse sections, selection of wall and plaque ROIs in longitudinal sections, estimation of texture features in different image areas, motion analysis of tissue ROIs, and clustering of the extracted features. It is concluded that Analysis can provide a useful platform for computerized analysis of medical images and support of diagnosis

Impact of local attenuation approximations when estimating correlation length from backscattered ultrasound echoes

Estimating the characteristic correlation length of tissue microstructure from the backscattered power spectrum could improve the diagnostic capability of medical ultrasound. Previously, size estimates were obtained after compensating for source focusing, the frequency-dependent attenuation along the propagation path (total attenuation), and the frequency-dependent attenuation in the scattering region (local attenuation). In this study, the impact of approximations of the local attenuation on the scatterer size estimate was determined using computer simulations and theoretical analysis. The simulations used Gaussian impedance distributions with an effective radius of 25 microm randomly positioned in a homogeneous half-space sonified by a spherically focused source (f/1 to f/4). The approximations of the local attenuation that were assessed neglected local attenuation (i.e., assume 0 dB/cm-MHz) neglected frequency dependence of the local attenuation, and assumed a finite frequency dependence (i.e., 0.5 dB/cm-MHz) independent of the true attenuation of the medium. Errors in the scatterer size estimate due to the local attenuation approximations increased with increasing window length, increasing true local attenuation and increasing f number. The most robust estimates were obtained when the local attenuation was approximated by a tissue-independent attenuation value that was greater than 70% of the largest attenuation expected in the tissue region of interest.

Quantitation and localization of blood‐to‐brain influx by magnetic resonance imaging and quantitative autoradiography in a model of transient focal ischemia

The ability of gadolinium-diethylenetriaminepentaacetic acid (Gd-DTPA) enhanced MRI to localize and quantitate blood-brain barrier (BBB) opening was evaluated against quantitative autoradiographic (QAR) imaging of (14)C-alpha-aminoisobutyric acid (AIB) distribution. The blood-to-brain transfer constant (K(i)) for Gd-DTPA was determined by MRI in rats after 3 h of focal cerebral ischemia plus 2.5 h of reperfusion (n = 9), and that of AIB was determined by QAR shortly thereafter. Tissue regions of interest (ROIs) for Gd-DTPA leakage were identified by ISODATA segmentation of pre- and post-Gd-DTPA Look-Locker (L-L) T(1) maps. Patlak plots were constructed using time course of blood and tissue T(1) changes induced by Gd for estimating K(i). Among the nine rats, 14 sizable regions of AIB uptake were found; 13 were also identified by ISODATA segmentation. Although the 13 MRI-ROIs spatially approximated those of AIB uptake, the segmentation sometimes missed small areas of lesser AIB uptake that did not extend through more than 60% of the 2.0-mm-thick slice. Mean K(i)'s of AIB were highly correlated with those of Gd-DTPA across the 13 regions; the group means (+/-SD) were similar for the two tracers (7.1 +/- 3.3 x 10(-3) and 6.8 +/- 3.5 x 10(-3) ml.g(-1) . min(-1), respectively). In most instances, Gd-DTPA MRI accurately localized areas of BBB opening.

Three-dimensional subzone-based reconstruction algorithm for MR elastography.

Accurate characterization of harmonic tissue motion for realistic tissue geometries and property distributions requires knowledge of the full three-dimensional displacement field because of the asymmetric nature of both the boundaries of the tissue domain and the location of internal mechanical heterogeneities. The implications of this for magnetic resonance elastography (MRE) are twofold. First, for MRE methods which require the measurement of a harmonic displacement field within the tissue region of interest, the presence of 3D motion effects reduces or eliminates the possibility that simpler, lower-dimensional motion field images will capture the true dynamics of the entire stimulated tissue. Second, MRE techniques that exploit model-based elastic property reconstruction methods will not be able to accurately match the observed displacements unless they are capable of accounting for 3D motion effects. These two factors are of key importance for MRE techniques based on linear elasticity models to reconstruct mechanical tissue property distributions in biological samples. This article demonstrates that 3D motion effects are present even in regular, symmetric phantom geometries and presents the development of a 3D reconstruction algorithm capable of discerning elastic property distributions in the presence of such effects. The algorithm allows for the accurate determination of tissue mechanical properties at resolutions equal to that of the MR displacement image in complex, asymmetric biological tissue geometries. Simulation studies in a realistic 3D breast geometry indicate that the process can accurately detect 1-cm diameter hard inclusions with 2.5x elasticity contrast to the surrounding tissue.

Measurement of interstitial albumin in human skeletal muscle and adipose tissue by open-flow microperfusion.

The absolute concentration of albumin was measured in the interstitial fluid of subcutaneous adipose tissue and skeletal muscle in six healthy volunteers by combining the method of open-flow microperfusion and the no-net-flux calibration technique. By use of open-flow microperfusion, four macroscopically perforated double lumen catheters were inserted into the tissue regions of interest and constantly perfused. Across the macroscopic perforations of the catheters interstitial fluid was partially recovered in the perfusion fluid. Catheters were perfused with five solutions, each containing different concentrations of albumin. Absolute interstitial albumin concentrations were calculated by applying linear regression analysis to perfusate vs. sampled albumin concentration (no-net-flux calibration technique). Interstitial albumin concentrations were significantly lower (P < 0.0001) in adipose tissue (7.36 g/l; r = 0.99, P < 0.0003; range: 4.3-10.7 g/l) and in skeletal muscle (13.25 g/l; r = 0.99, P < 0.0012; range: 9.7 to 15.7 g/l) compared with the serum concentration (48.9 +/- 0.7 g/l, mean +/- SE, n = 6; range: 46.4-50.4 g/l). Furthermore, interstitial albumin concentrations were significantly higher in skeletal muscle compared with adipose tissue (P < 0.01). The study indicates that open-flow microperfusion allows stable sampling of macromolecules from the interstitial space of peripheral tissue compartments. Moreover, the present data report for the first time in healthy humans in vivo the true albumin concentrations of interstitial fluid of adipose tissue and skeletal muscle.

Noninvasive quantification of the cerebral metabolic rate for glucose using positron emission tomography, 18F-fluoro-2-deoxyglucose, the Patlak method, and an image-derived input function.

The authors developed and tested a method for the noninvasive quantification of the cerebral metabolic rate for glucose (CMRglc) using positron emission tomography (PET), 18F-fluoro-2-deoxyglucose, the Patlak method, and an image-derived input function. Dynamic PET data acquired 12 to 48 seconds after rapid tracer injection were summed to identify carotid artery regions of interest (ROIs). The input function then was generated from the carotid artery ROIs. To correct spillover, the early summed image was superimposed over the last PET frame, a tissue ROI was drawn around the carotid arteries, and a tissue time activity curve (TAC) was generated. Three venous samples were drawn from the tracer injection site at a later time and used for the spillover and partial volume correction by non-negative least squares method. Twenty-six patient data sets were studied. It was found that the image-derived input function was comparable in shape and magnitude to the one obtained by arterial blood sampling. Moreover, no significant difference was found between CMRglc estimated by the Patlak method using either the arterial blood sampling data or the image-derived input function.

The effect of heart motion on kinetic parameter estimates for dynamic cardiac SPECT

Dynamic cardiac SPECT is used to estimate kinetic rate parameters that describe the washin and washout of tracer activity between the blood and the myocardial tissue. These kinetic parameters correlate to myocardial perfusion. There are, however, many physical aspects associated with dynamic SPECT which introduce bias and variance into the estimates. This paper describes a study which investigates the effect of heart motion on kinetic parameter estimates. Dynamic SPECT simulations are performed using a beating version of the MCAT phantom. The amount of blood and background activity in the myocardial tissue regions of interest is shown to vary over the cardiac cycle causing errors in the kinetic parameter estimates, particularly estimates of the washin rate constant. The effect of cardiac motion on kinetic parameter estimates is reduced by bias and variance introduced by photon noise and geometric collimator response. This suggests that techniques used to correct for motion must do so without further reducing photon statistics.

False-positive reduction technique for detection of masses on digital mammograms: Global and local multiresolution texture analysis

We investigated the application of multiresolution global and local texture features to reduce false-positive detection in a computerized mass detection program. One hundred and sixty-eight digitized mammograms were randomly and equally divided into training and test groups. From these mammograms, two datasets were formed. The first dataset (manual) contained four regions of interest (ROIs) selected manually from each of the mammograms. One of the four ROIs contained a biopsy-proven mass and the other three contained normal parenchyma, including dense, mixed dense/fatty, and fatty tissues. The second dataset (hybrid) contained the manually extracted mass ROIs, along with normal tissue ROIs extracted by an automated Density-Weighted Contrast Enhancement (DWCE) algorithm as false-positive detections. A wavelet transform was used to decompose an ROI into several scales. Global texture features were derived from the low-pass coefficients in the wavelet transformed images. Local texture features were calculated from the suspicious object and the peripheral subregions. Linear discriminant models using effective features selected from the global, local, or combined feature spaces were established to maximize the separation between masses and normal tissue. Receiver Operating Characteristic (ROC) analysis was conducted to evaluate the classifier performance. The classification accuracy using global features were comparable to that using local features. With both global and local features, the average area, Az, under the test ROC curve, reached 0.92 for the manual dataset and 0.96 for the hybrid dataset, demonstrating statistically significant improvement over those obtained with global or local features alone. The results indicated the effectiveness of the combined global and local features in the classification of masses and normal tissue for false-positive reduction.

Regional perfusion parameters from pulmonary microfocal angiograms.

An indicator-dilution model was developed to describe transport of vascular contrast medium through an organ during acquisition of vascular dynamic contrast images. The model provides the theoretical basis for methods of determining regional blood flow, blood volume, and mean transit time from time-absorbance curves acquired from the images of tissue regions of interest (ROI) distal from the inlet site. The robustness of these methods was evaluated using a computer-simulated vessel network, which simulated the passage of a bolus of contrast medium through arterioles, networks of capillaries, and venules. The network was used to evaluate the reliability of ROI parameter estimation methods when the underlying model assumptions are violated. The shape of the ROI inlet concentration curve and moderate amounts of random noise did not affect the ability of the method to recover accurate parameter estimates. The estimates of ROI flow and transit time were degraded in the presence of significant dispersion of the inlet concentration curve as it traveled through arteries upstream from the microvascular ROI or when the flow was redistributed within the ROI. The estimates of ROI volume were relatively robust. The method was applied to image data of the dog pulmonary vasculature obtained using microfocal X-ray angiography to show that the results obtained from the simulations are consistent with actual data.

Using linear time-invariant system theory to estimate kinetic parameters directly from projection measurements

It is common practice to estimate kinetic parameters from dynamically acquired tomographic data by first reconstructing a dynamic sequence of three-dimensional reconstructions and then fitting the parameters to time activity curves generated from the time-varying reconstructed images. However, in SPECT, the pharmaceutical distribution can change during the acquisition of a complete tomographic data set, which can bias the estimated kinetic parameters. It is hypothesized that more accurate estimates of the kinetic parameters can be obtained by fitting to the projection measurements instead of the reconstructed time sequence. Estimation from projections requires the knowledge of their relationship between the tissue regions of interest or voxels with particular kinetic parameters and the projection measurements, which results in a complicated nonlinear estimation problem with a series of exponential factors with multiplicative coefficients. A technique is presented in this paper where the exponential decay parameters are estimated separately using linear time-invariant system theory. Once the exponential factors are known, the coefficients of the exponentials can be estimated using linear estimation techniques. Computer simulations demonstrate that estimation of the kinetic parameters directly from the projections is more accurate than the estimation from the reconstructed images.

Review of MR image segmentation techniques using pattern recognition.

This paper has reviewed, with somewhat variable coverage, the nine MR image segmentation techniques itemized in Table II. A wide array of approaches have been discussed; each has its merits and drawbacks. We have also given pointers to other approaches not discussed in depth in this review. The methods reviewed fall roughly into four model groups: c-means, maximum likelihood, neural networks, and k-nearest neighbor rules. Both supervised and unsupervised schemes require human intervention to obtain clinically useful results in MR segmentation. Unsupervised techniques require somewhat less interaction on a per patient/image basis. Maximum likelihood techniques have had some success, but are very susceptible to the choice of training region, which may need to be chosen slice by slice for even one patient. Generally, techniques that must assume an underlying statistical distribution of the data (such as LML and UML) do not appear promising, since tissue regions of interest do not usually obey the distributional tendencies of probability density functions. The most promising supervised techniques reviewed seem to be FF/NN methods that allow hidden layers to be configured as examples are presented to the system. An example of a self-configuring network, FF/CC, was also discussed. The relatively simple k-nearest neighbor rule algorithms (hard and fuzzy) have also shown promise in the supervised category. Unsupervised techniques based upon fuzzy c-means clustering algorithms have also shown great promise in MR image segmentation. Several unsupervised connectionist techniques have recently been experimented with on MR images of the brain and have provided promising initial results. A pixel-intensity-based edge detection algorithm has recently been used to provide promising segmentations of the brain. This is also an unsupervised technique, older versions of which have been susceptible to oversegmenting the image because of the lack of clear boundaries between tissue types or finding uninteresting boundaries between slightly different types of the same tissue. To conclude, we offer some remarks about improving MR segmentation techniques. The better unsupervised techniques are too slow. Improving speed via parallelization and optimization will improve their competitiveness with, e.g., the k-nn rule, which is the fastest technique covered in this review. Another area for development is dynamic cluster validity. Unsupervised methods need better ways to specify and adjust c, the number of tissue classes found by the algorithm. Initialization is a third important area of research. Many of the schemes listed in Table II are sensitive to good initialization, both in terms of the parameters of the design, as well as operator selection of training data.(ABSTRACT TRUNCATED AT 400 WORDS)

Spectral analysis of dynamic PET studies.

We describe a new technique for the analysis of dynamic positron emission tomography (PET) studies in humans, where data consist of the time courses of label in tissue regions of interest and in arterial blood, following the administration of radiolabelled tracers. The technique produces a simple spectrum of the kinetic components which relate the tissue's response to the blood activity curve. From this summary of the kinetic components, the tissue's unit impulse response can be derived. The convolution of the arterial input function with the derived unit impulse response function gives the curve of best fit to the observed tissue data. The analysis makes no a priori assumptions regarding the number of compartments or components required to describe the time course of label in the tissue. Rather, it is based on a general linear model, presented here in a formulation compatible with its solution using standard computer algorithms. Its application is illustrated with reference to cerebral blood flow, glucose utilization, and ligand binding. The interpretation of the spectra, and of the tissue unit impulse response functions, are discussed in terms of vascular components, unidirectional clearance of tracer by the tissue, and reversible and irreversible phenomena. The significance of the number of components which can be identified within a given datum set is also discussed. The technique facilitates the interpretation of dynamic PET data and simplifies comparisons between regions and between subjects.