New Three-Dimensional Echocardiographic System Using Digital Radiofrequency Data—Visualization and Quantitative Analysis of Aortic Valve Dynamics With High Resolution
Methods, Feasibility, and Initial Clinical Experience
Background— Common 3D systems have only limited spatial and temporal resolution (frame rate of 25 Hz). Thin structures such as cardiac valves are not imaged exactly; rapid movement patterns cannot be precisely recorded. The objective of the present project was to achieve radiofrequency (RF) data transmission to the 3D workstation to improve image resolution.
Methods and Results— A commercially available echocardiographic system (5-MHz transesophageal echocardiography probe) with an integrated raw data interface enables transmission of RF data (up to 40 megabytes per second). A 3D data set may contain up to 3 gigabytes, so that all of the high-resolution ultrasound information of the 2D image is available. Frame rates of up to 168 Hz result in temporal resolution 6 times that of standard 3D systems. The applicability of the system and the image quality were tested in 10 patients. The structure of the aortic valve and the dynamic changes were depicted by volume rendering. The changes in the orifice areas were measured in frame-by-frame planimetry. The mean number of frames recorded per cardiac cycle was 122±16. The improved structural resolution enabled a detailed imaging of the morphology of the aortic cusps. The rapid systolic movement patterns were recorded with up to 51 frames. The high number of frames enabled creation of precise area-time diagrams. Thus, the individual phases of aortic valve movement (rapid opening, slow valve closing, and rapid valve closing) could be analyzed quantitatively.
Conclusion— A 3D system based on RF data enables high-resolution imaging of cardiac movement patterns. This offers new perspectives for qualitative and quantitative analyses, especially of cardiac valves.
Original received January 9, 2003; de novo received March 14, 2003; accepted May 6, 2003.
The impetus for the technical development of 3D echocardiography was its greater diagnostic potential than that of 2D examinations.1,2 The spatial imaging of cardiac structures is especially important in the planning and performance of surgical and catheter interventional procedures.3,4 In addition, new possibilities of quantitative analysis, such as the determination of ventricular size and function, are also offered.5,6 Establishing 3D echocardiography as a clinical routine procedure requires, however, further technical improvements in spatial and temporal image resolution.
The rationale for the technical realization of radiofrequency (RF) data transmission from the ultrasound unit to the 3D workstation via a raw data interface is the attendant considerable improvement in spatial and temporal image resolution. In this article, we present a system in which the frame rate has been increased 6 times to 168 Hz. Practical applicability and image quality of the system were tested under clinical conditions.
Data Acquisition Setup
The investigations were performed using a PowerVision 6000 ultrasound system (Toshiba Corp) equipped with a 5-MHz multiplane transesophageal echocardiography (TEE) probe and a digital RF data output. A 3D system with modified Echo-Scan software (TomTec GmbH) was used as a control unit. RF data were acquired at a data rate of 40 megabytes per second using a parallel input/output interface. RF data were directly transmitted to a data acquisition unit. Bidirectional serial command lines were used to automatically detect the ultrasound system settings as well as to control data acquisition from the remote computer and to acknowledge execution of commands (Figure 1).
Postprocessing and Volume Reconstruction
The acquired RF data correspond directly to the beam-formed backscatter signals received by the ultrasonic transducer. In a first step, image lines are created from the RF data by means of echo processor software. Subsequently, 2D ultrasonic images equal to those displayed on the ultrasound system monitor are generated using scan converter software. Each data set contains ∼130 frames (average heart rate, 80 bpm; frame rate, 168 frames per second). An angle increment of 5° was chosen to acquire the 3D data resulting in 36 scanning positions to cover the entire 180° rotation range of the multiplanar TEE probe. Individual images are created for each frame in all 36 angle positions, resulting in a total number of 4680 images per data set. To equalize the number of frames per angle position, the minimal number of frames over all angles is determined and applied to all other positions.
All data sets are stored in a file format complying with the standard file format normally used in the video acquisition setup. This enables us to apply the Echo-View software (TomTec) directly for volume reconstruction.
Despite the narrow sector angle (45°) used, each data set aggregates to ≈3 gigabytes. Applicable review data sets were generated by selecting a region of interest. Thus, the average data set size was reduced to ≈1.2 gigabytes.
The clinical applicability of the system was tested in 10 patients (3 women, 7 men; mean age, 54±13 years). The aortic valve was examined with the maximum frame rate (168 Hz). Multiplane data recording was performed in 5-degree increments. The study was approved by the local Ethics Committee. An individual examination was performed after the patient had granted informed consent.
A mean time of 6.5±0.5 minutes was required for the acquisition (5.8 to 7.5 minutes). Depending on the patient’s heart rate, between 102 and 149 frames per cardiac cycle could be recorded (mean, 122±16 frames).
Structural Image Resolution of the Aortic Valve
Figure 2A shows an aortic valve generated from video signal data. The edges of the cusps appear too thick as a result of the low structural resolution. The use of RF data, by contrast, leads to good structural image resolution in the 3D data set (Figure 2B). Compared with the 2D echocardiographic image, the cusp edges in the 3D anyplane mode are recorded with almost the same detail. This is a prerequisite for a high-resolution spatial image of the valve in the 3D volume-rendered mode.
Dynamic Changes of the Aortic Valve During the Cardiac Cycle
The systolic changes of an aortic valve are shown in Figure 3A in high temporal resolution. The valve opening begins with a separation of the cusps; then there is a very rapid increase in the orifice area. The maximum orifice is attained already in the early systole. Valve closing proceeds in 2 phases; after the maximum orifice has been attained, the valve begins a slow closing movement. Toward the end of the systole, there is a rapid valve closing movement. The course of the orifice area over time is shown in Figure 3B. The 3 phases of systolic opening are recorded in detail and can be quantitatively analyzed. The mean orifice area after the rapid opening was 2.80±1.05 cm2; at the end of the slow valve closing, 1.83±0.91 cm2. The mean velocities for rapid valve opening and for slow and rapid closing were 64.1±22.0, 5.4±3.8, and 57.6±33.1 cm2/s, respectively. The mean time required for planimetry of the orifice areas was 13±2 minutes.
Improvement in 3D echocardiographic image resolution is mandatory for expanded scientific and broad clinical applications. The present study demonstrates that the use of digital RF data improves the image quality and also brings decisive improvement in the possibility of quantitative analysis.
Current 3D Echocardiographic Imaging Techniques
Only low frame rates of ≈25 Hz can be achieved using commercially available 3D systems based on multiplane examination. In a further development of the multiplane technique, the temporal image resolution could be doubled to 50 Hz by modification of the acquisition software.
This opened new possibilities for 3D echocardiographic analysis of cardiac dynamics.7,8 An alternative procedure is 3D echocardiography with phased-array real-time volumetric systems (RT3DE).6 RT3DE enables rapid data acquisition during a cardiac cycle, and the latest developments also enable real-time generation of the 3D image. However, RT3DE, in addition to limited spatial resolution, offers only a poor temporal image resolution (≈20 Hz).
Clinical Applicability of the New System
The ultrasound unit used is a commercially available system, which was technically modified. A commercially available TEE probe can be used for 3D acquisition. This will facilitate integration into clinical routine. Because greater quantities of data are processed, the acquisition time is somewhat longer than in common systems, despite improved computer performance. The examinations were performed in small angle increments of 5 degrees, to obtain good structural resolution of the aortic valve. We required a mean time of 6 to 7 minutes for an acquisition. This is acceptable under clinical conditions.
New Perspectives for 3D Echocardiography
Real-time generation of 3D images and markedly improved image resolution are the 2 most important advances for broad clinical application of 3D echocardiography. The system that we are presenting is limited by the multiplane technique (longer acquisition time and offline analysis), but thanks to its good spatial and temporal image resolution, it opens new diagnostic possibilities for 3D echocardiography. It is therefore a measure for the requirements for future real-time systems. Compared with the usual systems, the transfer of RF data enables a considerably greater data flow to the 3D workstation. This means that complete 2D ultrasound information is available for 3D reconstruction. Thus, the morphology of cardiac structures can be reconstructed in more detail. Especially in thin structures such as cardiac valves, good structural resolution is decisive for the quality of the reconstruction and quantification of the orifice area.9 A high temporal resolution creates new possibilities for quantitative analyses. Knowledge of aortic valve function has been obtained primarily from experimental studies in animals, because no imaging procedure has been able to record rapid movement patterns in humans with sufficient accuracy.10,11 In a recent study (using a system with a 50-Hz image rate), we could show for the first time in addition to the normal function how aortic valve movement is influenced by myocardial and valvular factors.8 The further increase in frame rate to 168 Hz especially improves diagnosis with respect to the short phase of valve opening, which occurs at a high speed.
A new 3D system with a high frame rate can thus contribute to improved understanding of aortic valve function. Possible clinical applications are analyses of the function of aortic valve bioprostheses, diagnostics of valve function after valve-preserving surgery, or examinations of function and progression of stenosed aortic valves.
A 3D system based on RF data enables high-resolution imaging of cardiac movement patterns. This offers new perspectives for qualitative and quantitative analyses, especially of cardiac valves.
Supported by the Deutsche Forschungsgemeinschaft (German Research Foundation).
Movies are available in the online-only Data Supplement at http://www.circulationaha.org.
Wollschläger H, Zeiher AM, Klein HP, et al. Transesophageal echo computer tomography (ECHO-CT): a new method of dynamic 3-D reconstruction of the heart. Biomed Tech. 1989; 34: 10–11.
Buck T, Hunold P, Wentz KU, et al. Tomographic three-dimensional echocardiographic determination of chamber size and systolic function in patients with left ventricular aneurysm. Circulation. 1997; 96: 4286–4297.
Shiota T, Jones M, Chikada M, et al. Real-time three-dimensional echocardiography for determining right ventricular stroke volume in an animal model of chronic right ventricular volume overload. Circulation. 1998; 97: 1897–1900.
Handke M, Schäfer DM, Müller G, et al. Dynamic changes of atrial septal defect area: new insights by three-dimensional volume-rendered echocardiography with high temporal resolution. Eur J Echocardiogr. 2001; 2: 46–51.
Handke M, Heinrichs G, Beyersdorf F, et al. In vivo analysis of aortic valve dynamics by transesophageal 3D-echocardiography with high temporal resolution. J Thorac Cardiovasc Surg. 2003; 125: 1472–1479.