A digital 3-dimensional method for computing great artery flows: in vitro validation studies.

BACKGROUND: Conventional 2-dimensional Doppler large vessels are prone to inaccuracy. Three-dimensional (3D) volume imaging provides the opportunity to make cross-sectional flow calculations through digital spatiotemporal integration of flow velocity, area, and profile. METHODS: A new digital 3D color Doppler reconstruction method was used to generate radially acquired flow data sets. Raw scanline data with digital velocity assignments, obtained by scanning parallel to flow, were transferred from a specially programmed but otherwise conventional ultrasonographic system, which controlled a multiplane transesophageal probe, to a computer workstation via an Ethernet link for assimilation into color 3D data sets. This configuration was used to study 20 pulsatile laminar flows (stroke volumes 30 to 70 mL and peak flow rates 65 to 205 mL/s) in a curved tube model with an oval cross-sectional geometry. After generation of the color 3D data set, flow velocity values from cross sections perpendicular to the tubes were analyzed to determine flow rate and stroke volume. RESULTS: The flows from 3D digital velocity profiles showed close correlation with peak instantaneous flow rates (r = 0.99, y = 1.01x-0.9, standard error of estimate 4.1 mL/s). When interpreted with pulsed wave Doppler data obtained through the cardiac cycle, they also allowed computation of stroke volume (r = 0.98, y = 1.44x-2.5, standard error of estimate 3.8 mL). CONCLUSION: The ability to compute laminar flows from 3D digital data sets obtained parallel to the direction of flow and without the need for geometric assumptions represents an important opportunity for and advantage of 3D color Doppler echocardiography.

Direct computation of multiple 3D flow convergence isovelocity surfaces from digital 3D reconstruction of colour Doppler data of the flow convergence region: an in vitro study with differently shaped orifices.

AIMS: Evaluation of the accuracy of direct computation of multiple three-dimensional (3D) flow convergence (FC) isovelocities by using digital reconstruction of colour Doppler data of the flow convergence region. METHODS AND RESULTS: We used a conventional ultrasound system (ATL HDI 3000) connected to a computer workstation via Ethernet link. The digital 3D datasets were directly transferred to a Silicon Graphic Imaging Octane workstation for later measurement. We generated nine pulsatile flows (20-60 ml/beat), with peak flow rates (67-180 ml/s), through three orifices (circular, rectangular and triangular, S=0.24 cm(2)). The 3D reconstructions of FC surface areas from multi-threshold velocities, including aliasing velocities were analysed to quantify the peak flow rate. For all orifices, linear regression demonstrated excellent correlation between the 3D calculated and electromagnetic flow meter recorded data. While there was a high correlation for 3D computation of flow rate from the single best Nyquist of 24 cm/s (r=0.97-0.98, SEE=7 .75-12.58 ml/s), the ability to average three threshold velocities (15, 18 and 24 cm/s) yielded an improved correlation (r=0.98-0.99, SEE=5.70-7.73 ml/s). CONCLUSIONS: Direct computation of multiple 3D FC isovelocities from digital reconstruction of colour Doppler data of the FC region provides the potential to accurately quantify the complex asymmetric spatial flow events at any selected velocity.

Intragastric distribution and gastric emptying assessed by three-dimensional ultrasonography.

BACKGROUND & AIMS: Three-dimensional (3D) ultrasound imaging of the total stomach volume has not yet been achieved. The aim of this study was to investigate whether a magnetic position sensor system for acquisition of 3D ultrasonograms could be used to determine gastric emptying rates and intragastric distribution. METHODS: A system for position and orientation measurement was interfaced to an ultrasound scanner. In vitro accuracy was evaluated by scanning a porcine stomach. Fourteen volunteers, with a median age of 35 years, were scanned fasting and postcibally by two-dimensional (2D) and 3D ultrasound after ingesting a 500-mL soup meal. RESULTS: This 3D system yielded a strong correlation (r = 0.997) between true and estimated volumes in vitro. The limits of agreement were -9.1:70.1 mL in the volume range 1200-1900 mL. The intersubject variability of the total gastric volumes ranged from 12.5% to 46.0%, less than for antral area variability. The average half-emptying time was 22.1 +/- 3.8 minutes. Intragastric distribution of the meal, expressed as proximal distal volume, varied on average from 3.6 +/- 2.1 (5 minutes postpradially) to 2.7 +/- 1.9 (30 minutes postprandially). CONCLUSIONS: This 3D ultrasound system using magnetic scanhead tracking showed excellent in vitro accuracy, calculated gastric emptying rates more precisely than by 2D ultrasound, and enabled estimation of intragastric distribution of a soup meal.

Carotid plaque morphology and clinical events.

BACKGROUND AND PURPOSE: Studies have suggested that B-mode ultrasonography can be used to determine carotid plaque composition and that specific plaque characteristics are associated with a worse clinical outcome. However, histological studies examining the relationship between carotid plaque morphology and clinical outcome have reported conflicting findings. Furthermore, few investigators have described plaque morphology in quantifiable terms. This study examines the association between the volume of carotid plaque constituents and preoperative ischemic neurological symptoms. Constituents examined were chosen based on their potential for identification by current diagnostic imaging modalities such as ultrasound or MRI. METHODS: Atherosclerotic plaques from 43 patients undergoing carotid endarterectomy were examined histologically, with sections obtained every 0.5 to 1 mm. The lesions were examined for the presence and quantity of fibrous intimal tissue, intraplaque hemorrhage, lipid core, necrotic plaque core, and calcification. The quantity of each constituent was compared in plaques removed from symptomatic patients with those excised from asymptomatic individuals. Differences were analyzed with a Kolmogorov-Smirnov statistic. RESULTS: There was no difference between plaques removed from asymptomatic and symptomatic patients with regard to the presence and volume of fibrous intimal tissue, intraplaque hemorrhage, the lipid core, the necrotic core, or calcification. CONCLUSIONS: In patients with highly stenotic carotid lesions who are undergoing carotid endarterectomy, gross plaque composition is similar regardless of preoperative symptom status. Given this similarity, it is unlikely that differences in the volume of intraplaque hemorrhage, lipid core, necrotic core, or calcification in atherosclerotic carotid plaques explain their embolic history.

Performance of a miniature magnetic position sensor for three-dimensional ultrasound imaging.

A miniature magnetic position sensor used for three-dimensional ultrasound imaging was tested for precision and accuracy in vitro. The sensor alone was able to locate points with root-mean-square (rms) uncertainty of 1.7 mm and accuracy of 0.05 +/- 0.62 mm over its specified operating range of 50 cm. With an ultrasound imaging system, a point was located from arbitrary viewing windows with 2.4-mm rms uncertainty and 0.06 +/- 0.68 mm accuracy. If viewing windows were limited to those representative of a typical ultrasound examination, the system could achieve rms uncertainty in point location of < 1 mm. Performance was not affected by operation of the imaging system when the sensor was mounted on an ultrasound scanhead. Sensitivity to metals in the operating environment was also measured.

A new method for studying plaque morphology.

A histologic method was developed for three-dimensional (3-D) analysis of atherosclerotic plaques removed from the carotid bifurcation during endarterectomy. By sectioning the plaque at frequent intervals (0.5 to 1.0 mm), it is possible to obtain important information on plaque constituents with regard to their volume and distribution within the lesion. These data from each section are combined with those from other sections and displayed in a 3-D format for the entire length of the lesion. The tissues making up each of the 10 carotid plaques were outlined and digitized for each histologic section by position along the lesion. From the areas outlined a 3-D model was created by a computer-aided design program. Quantitative information on tissue distribution within the plaque was measured. Fibrous tissue constituted between 35% and 70% of plaque volume; loose necrosis from 0.5% to 30% of the plaque and thrombus occupied, at a maximum, 10% even though if was present in six of the 10 plaques. To investigate the distribution of constituents about the long axis, measurements were also made from each of the four quadrants of each section. The reproducibility of the measurements of three sets of sections at 10-mm separation showed that estimates of the amount of some constituents were very reproducible whereas others had considerable variation related to the small volume they occupied within the lesion. By generating a complete 3-D reproduction of the contents of atherosclerotic plaques, it may be possible to identify those features of the plaque that are most responsible for the development of ischemic events.

Hemodynamics of stenotic infrainguinal vein grafts: theoretic considerations.

We developed a theoretic model of arterial stenosis to study the relationship between perfusion pressure and regional hemodynamics in stenotic infrainguinal vein grafts in an attempt to identify grafts at high risk for failure. Our model was based on the concept of energy and mass conservation of the flowing blood. We used the modified Bernoulli equation (delta P = 4 delta V2) to calculate the maximum possible intrastenotic peak systolic velocity (PSV) from the systolic blood pressure. PSV was measured by means of duplex ultrasonography in infrainguinal bypasses up to the time of revision (nine grafts) or spontaneous thrombosis (two grafts). We related arm systolic blood pressure, intrastenotic PSV, and prestenotic PSV obtained from duplex examinations conducted prior to graft thrombosis or revision and applied our model to these stenotic vein grafts. Intrastenotic PSV was consistently lower than maximum PSV predicted from the Bernoulli equation. The highest measured intrastenotic PSV of 600 cm/sec would require a minimum perfusion pressure of 144 mm Hg. The lowest measured PSV (20 cm/sec) was considered the minimum "thrombotic threshold velocity." This model predicts that for parabolic profile flow in an 80% diameter-reducing axisymmetric stenosis (96% cross-sectional area reduction), a prestenotic PSV of 20 cm/sec would produce an intrastenotic PSV of 500 cm/sec requiring the equivalent potential energy of 100 mm Hg systolic blood pressure. Our theory implies that in patients with nocturnal hypotension thrombosis of stenotic vein grafts may occur.

Ultrasound determination of total arterial wall thickness.

PURPOSE: Ultrasonic measurement techniques for determining intima-media thickness and total arterial wall thickness have been described. The intima-media thickness measurements are currently in use in large epidemiologic trials. Intima-media thickness does not evaluate extramedial atherosclerotic change and so may not fully reflect pathologic changes in the arterial wall. METHODS: After we performed variability studies of B-mode image acquisition and measurement, we measured total wall thickness and intima-media thickness of the common carotid arteries in 60 adult subjects in three groups: a control group aged 20 to 29 years, a control group aged 60 to 79 years, and a claudication group aged 60 to 79 years. Measurements were made with B-mode ultrasound images. RESULTS: No statistical difference between sexes was noted. A statistically significant (p < or = 0.05) increase in intima-media thickness and wall thickness was found with increasing age, and an additional increase was observed with clinically significant lower extremity arterial occlusive disease (p < or = 0.05). Image quality had an effect on measurement accuracy. CONCLUSIONS: The finding that the wall thickness of common carotid arteries is increased in those patients with clinically significant lower extremity disease supports the theory that atherosclerosis affects the arterial wall in a systemic fashion. Because intima-media thickness also increases across subject groups without change in its proportional contribution to the total arterial wall thickness, extramedial arterial changes also occur with aging and the development of atherosclerosis. We propose that because increases in wall thickness measurements of common carotid arteries follow intima-media thickness increase (but do not necessarily measure the same physiologic change) and the wall thickness method can be used in cases when the intima-media thickness cannot be measured, arterial wall thickness measurement may serve as an alternate or confirmatory test of peripheral artery atherosclerotic severity.

Ultrasonic three-dimensional reconstruction: in vitro and in vivo volume and area measurement.

This study validates the use of an ultrasound three-dimensional reconstruction system to measure phantom and blood conduit geometry. Independently determined uniform and stenotic phantom dimensions are compared with reconstruction-based measurements. Lower extremity saphenous vein bypass graft reconstructions were performed to demonstrate clinical application. Uniform phantom independent and reconstructed volume correlation was high (r = 0.989), the average volume difference was 4.68 mm3 and the average area difference was 0.4 mm2. An in vitro 28% diameter reduction was detected. Stenotic bypass graft segment volume was 795 mm3; following successful angioplasty the volume increased to 1419 mm3. Advantages of this technique are its accuracy, the luminal information it provides and the absence of mechanical arm or acoustic transmitter limitations. We are exploring the possibility that measurement of luminal change over time may allow stenosis detection prior to hemodynamic disturbance, in an ongoing clinical saphenous vein bypass graft surveillance study.

Echolucent regions in carotid plaque: preliminary analysis comparing three-dimensional histologic reconstructions to sonographic findings.

This study compares sonographic and histologic findings within defined spatial regions in carotid artery plaque, using computer generated three-dimensional reconstructions. Twenty-four patients (14 asymptomatic, 10 symptomatic) with angiographically documented 70% to 99% carotid artery stenosis were examined with ultrasonic B-mode imaging prior to endarterectomy. Using a standardized protocol for instrument set-up and scanning technique, echolucent regions in the plaque were identified. After endarterectomy, each plaque was sectioned at 0.5 to 1.0 millimeter increments throughout its length. Sites containing intraplaque hemorrhage, cholesterol clefts, foam cells, necrotic cores, dense calcification and speckled calcification were identified. These areas were outlined on a template, digitized and imported into a computer program that created three-dimensional reconstructions of the histologic findings. Each carotid plaque was divided into quadrants for analysis: (1) lateral wall proximal to the common carotid bifurcation (flow divider); (2) medial wall proximal to the flow divider; (3) lateral wall distal to the flow divider; and (4) medial wall distal to the flow divider. The odds of finding intraplaque hemorrhage, foam cells, necrotic cores and speckled calcification were significantly higher in quadrants with an echolucent region identified by ultrasonography (odds ratio (95% confidence interval) for intraplaque hemorrhage = 3.5 (1.4-8.6); foam cells = 4.0 (1.6-9.9); necrotic cores = 3.2 (1.2-8.4); speckled calcification = 4.0 (1.6-9.8). This preliminary analysis demonstrates the potential of these newly developed techniques for comparing ultrasonic imaging to histology.

3D ultrasonic image feature localization based on magnetic scanhead tracking: in vitro calibration and validation.

The basis of a three-dimensional (3D) ultrasound imaging system was constructed from a commercially available magnetometer-based position and orientation measurement (POM) device, a standard B-Mode ultrasound instrument and a personal computer. To evaluate the system's performance, a novel method was devised using an iterative, least-squares technique to simultaneously determine the system's calibration parameters and measure its precision in locating points in three-dimensional space. When tested separately, the POM system located single points with a root mean squared (RMS) uncertainty of from 1.4 mm to 3.2 mm over the 610 mm working radius of the system. When combined with the ultrasound instrument, the RMS uncertainty in locating point targets varied from 2.1 mm to 3.5 mm. These results establish the lower limits of variability to be expected from this system when locating fiducial anatomical landmarks for repeated examinations of the same region of the body, and when making numerical 3D reconstructions from multiple planar images.

Three-dimensional transesophageal echocardiography for depiction of regional left-ventricular performance: initial results and future directions.

To assess the potential of a prototype transesophageal echocardiography probe for evaluating left-ventricular wall motion in three dimensions, we acquired images under anesthesia in 15 patients who had akinesia or dyskinesia and 8 patients who had normal function demonstrated on preoperative ventriculography. Short-axis, oblique transgastric scans were obtained in 16 of the patients and four-chamber, long-axis oblique scans were obtained in 12 patients, with five patients (22%) yielding good-quality scans of both types. Off-line, we outlined the endocardial borders manually and used the outlines to make computer-generated three-dimensional models of the endocardial surfaces, color-tiled according to regional ejection fraction. Compared with contrast ventriculograms, the regional ejection fraction histograms derived from these models showed 86% concordance for detecting dyssynergy. However, the concordance between the ventriculograms and the color-tiled models in localizing the dyssynergy was only 67% overall. Uncertainty in rotational alignment between the reconstructions and the ventriculograms appeared to contribute to misreading the location of dyssynergy. In addition, the apical region appeared to have been missed in 8 (50%) of the short-axis scans, whereas it was visualized in all long-axis scans. We conclude that three-dimensional analysis of the location, extent, and degree of left-ventricular dyssynergy is feasible from transesophageal echocardiograms and could have wide application in the study of regional ventricular function. However, improvements are necessary to enable the transducer to scan the cardiac apex more reliably from the short-axis viewpoint and to have a means for spatially orienting the images with respect to an external frame of reference.

Carotid artery intraplaque hemorrhage and stenotic velocity.

BACKGROUND AND PURPOSE: One of the proposed mechanisms for sudden expansion of a carotid bifurcation plaque is hemorrhage within the lesion. It has been postulated that the sudden increase in plaque size will acutely reduce blood flow to the ipsilateral hemisphere and induce either a transient ischemic attack or a stroke. In this study, the relation between peak systolic velocity at the site of narrowing and its potential role in the development of intraplaque hemorrhage were investigated. METHODS: Ten patients who had carotid endarterectomy were examined by duplex Doppler sonography before surgery to determine the peak systolic velocity at the site of maximal narrowing. The excised carotid plaques were sectioned at 1-mm intervals and examined for histological evidence of intraplaque hemorrhage. The recorded peak systolic velocities in patients with intraplaque hemorrhage were compared with the velocities in cases in which no hemorrhage was identified. RESULTS: Five of the ten patients had intraplaque hemorrhage. Four of the five patients with intraplaque hemorrhage had a peak systolic velocity of > 420 cm/sec and diastolic velocities of > 160 cm/sec; none of the patients without intraplaque hemorrhage had such high values. CONCLUSIONS: Peak systolic velocity is significantly higher in patients with intraplaque hemorrhage. The specificity and sensitivity of a peak systolic velocity of > 420 cm/sec in predicting intraplaque hemorrhage remains to be determined.

Left ventricular ejection fraction: single-plane and multiplanar transesophageal echocardiography versus equilibrium gated-pool scintigraphy.

The relative accuracy and precision of estimating left ventricular ejection fraction (EF) in dogs were assessed by two-dimensional transesophageal echocardiography (2D-TEE) and by three-dimensional transesophageal echocardiographic (3D-TEE) imaging and reconstruction. This assessment was accomplished by comparing each echocardiographic method to a gated equilibrium blood pool radionuclide (RN) standard. By using both correlation and regression analysis, 2D-TEE performed reasonably well in estimating RNEF (correlation coefficient [r] = 0.80, slope = 1.01, intercept = 6.37, standard error of the estimate [SEE], 8.98), but not as well as 3D-TEE (r = 0.86, slope = 0.83, intercept = 3.38, SEE, 5.74). Using Altman and Bland's methods of comparison analysis, it was found that 2D-TEE overestimated RNEF by 7% (standard deviation [SD], 8.8). This degree of overestimation was not consistent across the range of measurement. In contrast, 3D-TEE slightly underestimated RNEF by less than 3% and showed less variability (SD, 6.0). The accuracy of the 3D-TEE determinations was not dependent on the magnitude of EF. Additionally, a significantly higher proportion of the 2D-TEE measurements (0.30) compared with the 3D-TEE measurements (0.10) differed from RN values by more than 10% (P = 0.009, McNemar's test). At the clinically important low end of the EF range (RNEF less than or equal to 35%), 2D-TEE may be expected (with 95% confidence) to be within -15% to +28% EF of reference values, whereas 3D-TEE can be expected to be within -8% to +5% EF relative to RN.(ABSTRACT TRUNCATED AT 250 WORDS)

Ventricular volume measurement from a multiplanar transesophageal ultrasonic imaging system: an in vitro study.

We have developed a system to assess the feasibility of using multiple transesophageal ultrasonic images to measure left-ventricular volume, an important variable in patient management. The system includes a special transesophageal probe with a micromanipulator for acquiring cardiac images in multiple planes with known interplanar spatial relationship and an off-line processing system to compute the volume. In vitro studies with the probe demonstrated that the distance between two targets in space can be identified within 2 mm (SD = 0.4 mm) for points in the imaging plane 3.4 mm (SD = 0.5 mm) for points not lying in the imaging plane. This gives an average accuracy of +/- 6.5% for distances greater than 4.5 cm. Comparison of ultrasonic measurements of the volume of water-filled balloons and excised hearts to the volume required to fill them, revealed a correlation coefficient of 0.992, a regression line having a slope of 1.0 and an ordinate intercept at 0.2 mL, and a standard error of the estimate of 8 mL.

Matched filter identification of left-ventricular endocardial borders in transesophageal echocardiograms.

The use of one-dimensional spatial matched filtering for identifying the left ventricular endocardial borders in human transesophageal echocardiograms recorded during surgery is investigated. A maximum-likelihood method was used to choose the endocardial intensity profiles centered within the ventricle. The computer-generated border points were compared to those identified by an experienced ultrasonographer. A 16-pixel step template located 63.2% of the border points within 2 mm of the manual border. Median prefiltering of the images reduced detection accuracy by 3% to 6%. No statistically significant difference in accuracy was found between longer and shorter templates or between data-derived and step templates. Compared to manual estimates, computer generated cross-sectional area determinations were correlated with a coefficient of 0.93. Matched filtering executes rapidly, does not require prefiltering, and performs as well as other reported methods in estimating ventricular area.