Spatio-temporal image correlation (STIC): new technology for evaluation of the fetal heart

Authors


Abstract

Spatio-temporal image correlation (STIC) is a new approach for clinical assessment of the fetal heart. It offers an easy to use technique to acquire data from the fetal heart and to aid in visualization with both two-dimensional and three-dimensional (3D) cine sequences. The acquisition is performed in two steps: first, images are acquired by a single, automatic volume sweep. Second, the system analyzes the image data according to their spatial and temporal domain and processes an online dynamic 3D image sequence that is displayed in a multiplanar reformatted cross-sectional display and/or a surface rendered display. The examiner can navigate within the heart, re-slice, and produce all of the standard image planes necessary for a comprehensive diagnosis. The advantages of STIC for use in evaluation of the fetal heart are as follows: the technique delivers a temporal resolution which corresponds to a B-mode frame rate of approximately 80 frames/s; it provides the examiner with an unlimited number of images for review; it allows for correlation between image planes that are perpendicular to the main image acquisition plane; it may shorten the evaluation time when complex heart defects are suspected; it enables the reconstruction of a 3D rendered image that contains depth and volume which may provide additional information that is not available from the thin multiplanar image slices (e.g. for pulmonary veins, septal thickness); it lends itself to storage and review of volume data by the examiner or by experts at a remote site; it provides the examiner with the ability to review all images in a looped cine sequence. Copyright © 2003 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

In the mid 1980s the concept of the four-chamber view for screening of the fetal heart was introduced to obstetric scanning as a novel approach for the detection of congenital heart malformations1, 2. Following introduction of this concept, several studies reported poor performance of the second-trimester four-chamber screening examination3, 4. The RADIUS study found that ultrasound performed even at tertiary centers only identified 22.7% of fetuses with congenital heart defects3. Investigators subsequently realized that one of the limitations of the four-chamber view screening examination was that the outflow tracts were not being evaluated, resulting in minimal or no detection of tetralogy of Fallot, mild or moderate aortic or pulmonary stenosis, and transposition of the great arteries5, 6. Nevertheless, in a recent study in which the four-chamber and outflow tracts were examined, Stoll et al.4 found that only 13.7% of isolated congenital heart defects were detected during the second trimester in over 92 000 screened pregnancies.

Unlike the pediatric or adult patient, for whom pre-determined windows are used to obtain standardized views of the heart, the fetus presents several problems: (1) the size of the heart varies as a function of gestational age; (2) the fetal position may vary during the examination, making imaging of structures more difficult; (3) fetal movement during the examination may result in not acquiring all of the diagnostic images for analysis; (4) oligohydramnios may make imaging more difficult; (5) interpretation must be done in real time, either during the examination or after reviewing a video tape; (6) acquiring the images of the heart can be time consuming7, 8. Given these potential obstacles, evaluation of the fetal heart is one of the most challenging diagnostic tasks the fetal examiner encounters during a second- or third-trimester ultrasound examination.

Recently, investigators have reported using volume data acquisitions to reconstruct the anatomy of the fetal heart using either a static volume sweep or gated technology in which reconstruction of cardiac volumes is accomplished9–15. However, other than the static volume sweep, online dynamic three-dimensional (3D) image sequence volume acquisition and display of fetal cardiac anatomy reported in previous studies had the following limitations: it requires additional software and hardware that are not part of the ultrasound system; the image resolution may be suboptimal for second-trimester diagnosis; additional time is required for rendering. The purpose of this paper is to describe a new technological development, using online dynamic 3D image sequences, that addresses many of the above issues.

Ultrasound Equipment

The ultrasound system used in this study was a VOLUSON 730 Expert series (GE Medical Systems, Kretz Ultrasound, Zipf, Austria). The transducers capable of the STIC technique are the real-time probe (RAB 4–8 MHz) used for obstetric/gynecological and abdominal applications, and the vaginal real-time probe (RIC 5–9 MHz). STIC is an integrated part of the system's 3D/4D basic software option, requiring no additional software packages for online analysis. A software package was used for offline analysis which duplicates the online environment (GE Medical Systems).

The Technology: Descriptionand Limitations

STIC stores complete volume data sets of the fetal heart and reduces dependency on the examiner's experience for scanning cardiovascular anatomy. Recorded images are not only two-dimensional (2D) frames of the region of interest, but all of the information adjacent to it. As a consequence, the user can digitally store the acquired data, re-optimize and re-slice views of the heart during the clinical examination, or re-create the examination later. Additionally, data files may be transferred to or from remote sites for review by an expert in fetal cardiology.

Acquisition of the STIC volume

STIC is an automated volume acquisition in which the array inside the transducer housing performs a slow, single sweep, recording one single 3D data set. This volume consists of a high number of 2D frames, one behind the other. Due to the small region of interest required to image the fetal heart, the B-mode frame rate during the acquisition of the volume scan is very high, in the range of 150 frames/s. Assuming a volume acquisition time of 10 s and sweeping over an area of 25° (both parameters can be adjusted), there are 1500 B-mode images in the volume memory. During this acquisition time the fetal heart beats 20–25 times, which means there are 20–25 images showing a systolic peak contained within these 1500 B-mode frames (Figure 1). It is essential that the selected box size is as small as possible so that the fetal heart covers the major part of the image to be acquired, such that the relationship between moving and non-moving parts within the image covers between 30 and 50% of the overall image.

Figure 1.

Raw data volume showing a beating heart during a slow STIC sweep. The arrows indicate three continuous systolic cycles which are used to calculate the fetal heart rate.

Detection of the heart rate

When examining the atria and ventricles simultaneously, they are moving in opposite directions i.e. when the atria contract, the ventricles dilate and vice versa. STIC does not identify specific cardiac structures but analyzes the rhythmic movement, independent of its direction, and derives the heart rate from the periodicity of these movements.

Computation of beat-to-beat changes in the fetal heart rate

Assuming that the fetal heart does not beat at an absolutely precise rate, the STIC algorithm is required to detect changes in the heart rate. The algorithm is designed to identify minor changes in the heart rate that are evident but beyond the examiner's ability to detect them visually. Initially STIC calculates an average heart rate across all the acquired heartbeats. After the average heart rate is computed, beat-to-beat changes with a variability of about ±10% are detected. Considering these variations, the 2D images are rearranged accordingly, to make sure that images only from the same time during the heart cycle are merged into one volume. Whenever there are severe changes in the fetal heart rate during the STIC acquisition, the STIC algorithm has difficulties in calculating the average heart rate correctly, which can cause artifacts due to a rearrangement of non-corresponding images. From our experience it is rather unlikely that a major change of heart rate occurs during the short acquisition period of 7.5–15 s. If the examiner encounters this problem, a repeat acquisition can be obtained immediately.

Visualization of 2D images

The cardiac volumes are displayed as one real-time cardiac cycle, played in a cine loop. The loop may also be played in slow motion or stopped at any time for detailed analysis of specific phases of the cardiac cycle. Because there is a volume dataset, each of the scan planes can be moved and rotated while maintaining the synchronized cardiac loop. Thus, the four-chamber view, long axis, short axis, great vessels and all other views can be displayed both in cine loop or as still images. Different display formats are available: multiplanar (showing three planes perpendicular to each other), and single-plane views (Figure 2).

Figure 2.

Display formats using STIC. (a) Four simultaneous image displays: (A) demonstrates the plane of the acquired image obtained during the STIC sweep; (B) is the plane perpendicular to the white circle in plane (A) which would be in a line drawn from 6 o'clock to 12 o'clock through the point in the circle; (C) is the plane perpendicular to the white circle in plane (A) which would be in a line drawn between 9 o'clock and 3 o'clock through the point in the circle; (D) illustrates the three planes (A–C) intersecting the reference point within the white circle. The examiner can place the reference point within the white circle at any location in planes (A), (B) or (C) and observe the corresponding planes change their respective images. (b) An enlarged image of plane (A).

Visualization of the 3D rendered volume

From the volume data set a rendered 3D image of the heart can be created. Compared to the 2D image, in which a thin slice of anatomy is displayed, the rendered 3D image contains depth in the z-plane. The depth of the image can be controlled: with increased depth, the back wall of the chambers of the heart can be viewed while simultaneously viewing the atrioventricular valves (Figure 3); with decreased depth, a slice through the heart can be isolated (Figure 4). The rendered view can be examined as a cine loop or as a still image.

Figure 3.

Three-dimensional rendered four-chamber view in which the depth of the rendering has been increased to include the walls and the posterior wall of the heart. Images with (a) and without (b) labels showing the back walls of the atrial and ventricular chambers as well as the opening of the pulmonary vein (PV) within the left atrium (LA) and the inferior vena cava (IVC) in the right atrium (RA). LV, left ventricle; RV, right ventricle.

Figure 4.

Three-dimensional rendered four-chamber view in which the depth has been decreased to isolate the mid-portion of the heart. Images with (a) and without (b) labels in which the four-chamber view has been rotated to show the thickness of the ventricular and atrial septa (white arrows). LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle.

Imaging artifacts using STIC of the fetal heart

During image acquisition (7.5–15 s) fetal breathing or body movements will alter the images in the B and C planes (Figure 5). Excessive movement makes the interpretation of the images in these planes difficult, but minimal movement does not alter the anatomical relationships required for diagnosis. Breathing or movement of the fetus does not alter the A plane image, unless it is excessive. The reason for this is that the A plane contains the original, acquired B-mode images used in the construction of the B and C plane images.

Figure 5.

(a) During the STIC acquisition, which results in the (A) plane image, there was no fetal movement, as reflected by the constructed images in (B) and (C). (b) During the STIC acquisition fetal breathing occurred, resulting in undulations in the wall motion of images (B) and (C). Although fetal breathing alters these images, image (A) is still interpretable if the examiner evaluates the heart using only image (A) and moves through the volume from the beginning to the end of the acquisition without rotating the image, as was done in Figure 9.

Gestational age

We used STIC technology to successfully image the fetal heart between 13 and 40 weeks of gestation.

Potential clinical benefits using STIC for evaluation of the fetal heart

Improved temporal resolution

During the traditional 2D examination of the fetal heart the gray-scale display may be optimized using pre- and post-processing settings as well as harmonics to improve recognition of anatomical structures. Because of the increased number of frames acquired for a specific anatomical region using the STIC technique, this method results in improved temporal resolution of the online dynamic 3D image sequence that is displayed in the multiplanar or the surface rendered display. As a result, a second-trimester fetal heart that is difficult to image using 2D real-time technology becomes a better image when viewed in the STIC mode resulting from an increase in the number of 2D images stored in the volume data set (Figure 6).

Figure 6.

Hypoplastic left ventricle with a ventricular septal defect at 17 weeks of gestation. (a) Labeled and unlabeled four-chamber views recorded from the STIC acquisition and displayed in B-mode. (b) Rendered 3D image. Notice that the rendered image provides more detail and depth perspective as it relates to the hypertrophied left ventricular wall and septum (red overlay). In addition, the boundaries of the right atrial and ventricular walls are more pronounced in the rendered image. Ao, thoracic aorta; HLV, hypoplastic left ventricle; LA, left atrium; MV, mitral valve; RA, right atrium; RV, right ventricle; TV, tricuspid valve; VSD, ventricular septal defect.

Evaluation of the four-chamber view

During 2D real-time examination of the four-chamber view the examiner may focus on only one plane at a time. However, using STIC the examiner can evaluate the heart from the inferior or back walls to the superior parts of the four-chamber view (Figure 7). Using this technique ventricular and atrial septal defects may be identified that would otherwise be missed if only one plane of the four-chamber view was examined. In addition to moving through the inferior and superior planes, the examiner can rotate the four-chamber view and examine the 3D multiplanar anatomy in numerous planes around a 360° axis (Figure 8). For pediatric cardiologists this may be beneficial by allowing them to view the heart in imaging planes that are standard for the neonatal examination16–19. In addition to being able to rotate the image of the heart, end-diastolic and end-systolic views can be identified by evaluating the volume dataset frame-by-frame (Figure 9). The 3D rendered four-chamber view allows the examiner to evaluate depth relationships that may be displayed as a thicker slice of the four-chamber view, or include the full thickness of the cardiac chambers in which the posterior wall can be identified (Figures 3 and 4).

Figure 7.

Examination of different levels of the four-chamber view by manipulation of one STIC data set. The left panels give the labeled image, the right panels the original image. (a) At the level just superior to the inferior, diaphragmatic aspect. The interatrial septum is contiguous with the ventricular septum. (b) At the level of the foramen ovale. (c) At the level where the ascending aorta originates from the upper portion of the left ventricular chamber. LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle; FO, foramen ovale; PV, pulmonary vein; IA, interatrial septum; Ao, thoracic aorta; AA, ascending aorta.

Figure 8.

Planes of rotation of the four-chamber view which can be obtained using the software on the ultrasound machine. This allows the examiner to orient the views of the heart to comply with specific image planes that may be helpful in interpreting fetal cardiac anatomy.

Figure 9.

Using STIC, end-diastole (a) and end-systole (b) can be accurately determined because the single cardiac cycle contains image frames representing each of these phases of the cardiac cycle.

Visualization of the outflow tracts

Several techniques have been described to assist the fetal examiner in evaluation of the aortic and pulmonary outflow tracts using the four-chamber view as the reference point20–22. During live 2D real-time examination the outflow tracts may be difficult to image because of fetal orientation and movement. In addition, each examiner may have specific reference planes that they prefer to determine the relationships of the outflow tracts with their respective ventricles. Using STIC, coupled with the ability to rotate the data volume around a single reference point, makes assessment of the outflow tracts quite easy. Figure 10 illustrates a sequence of images demonstrating this principle.

Figure 10.

Rotation of the heart using STIC to identify the outflow tracts. (a,b) The five-chamber view in which the right pulmonary artery (a; straight arrow) is identified by the cross-hairs. Once the cursor is placed within the right main pulmonary artery, its corresponding bifurcation is visualized by rotating the image (curved arrow) until the pulmonary artery bifurcation is identified (b). (c,d) Identification of the aortic arch by placing the cross-hairs over the ascending aorta (c; straight arrow) and rotating the image (curved arrow) until the aortic arch is displayed. AA, ascending aorta; AArch, aortic arch; LPA, left pulmonary artery; RPA, right pulmonary artery.

Examination of complex heart defects

When a complex heart defect is encountered by the clinician, accurate diagnosis is important for appropriate prenatal counseling. When such a defect is encountered, it may take several hours to acquire all of the images to make an accurate diagnosis. The reason for this is that each part of the heart and outflow tracts must be examined individually as well as their relationships with other cardiac structures. Using STIC, the examiner only needs to acquire several volumes and then evaluate the individual structures of the heart. From our experience, STIC may decrease the evaluation time and improve our ability to identify complex intracardiac relationships (Figure 6).

Offline analysis

In the past offline analysis of the fetal heart required review of still images, videotape or digital files of the examination of the heart. The examiner did not have the ability to recreate the examination and had to rely only on the acquired images. The STIC volume allows a physician in one geographical area to send the study to a consultant fetal echocardiographer, irrespective of where they live. This opens up new opportunities for consultation and improved clinical diagnoses.

Patient communication

The improved ability to describe to the patient the fetal cardiac anatomy, especially when a malformation is suspected, may allow the patient to better understand the findings of the examination.

Conclusions

Unlike the adult or pediatric patient, for whom specific imaging windows are available for a systematic echocardiographic examination, the fetus can be in a number of positions and can move during the examination. Coupled with its small size, this may make the examination more difficult. With the introduction of STIC, data can be acquired within 7.5–15 s. Once the volume data is processed and displayed, examination of cardiovascular structures can be accomplished by rotating 2D images in an infinite number of planes. In addition, 3D surface rendered anatomy of the heart and great vessels which has the appearance of depth and volume can be accomplished. From our experience, using STIC technology has the following advantages:

  • 1.The technique delivers a temporal resolution which corresponds to a B-mode frame rate of approximately 80 frames/s when viewed in the cine-loop format.
  • 2.It provides the examiner with an unlimited number of images for review.
  • 3.It allows for correlation between image planes that are perpendicular to the main image acquisition plane.
  • 4.It has the potential to shorten the evaluation time, especially when complex heart defects are suspected.
  • 5.It enables the reconstruction of a 3D rendered image that contains depth and volume which may provide additional information that is not available from the thin 2D image slices.
  • 6.The data volume can be stored and reviewed offline by the examiner or it can be viewed by experts at a remote site.
  • 7.It provides the examiner with the ability to review all images in a cine-loop format.

It seems reasonable to conclude that if the concepts and techniques described here can be adequately used by more sonographers and sonologists, the antenatal diagnosis of congenital heart defects may be improved.

Ancillary