Time and Volume Curves

The software for LV analysis produces plots of the absolute segmental volume against time as well as normalized volume curves. When the absolute volumes are normalized such that the greatest volume represents 100%, the difference between 100% and the average of the minimum segmental volumes represents the EF. Segmental volume curves may be viewed individually, but the pattern of segmental contraction is useful and forms the basis of the SDI. The minimum volume point is identified on each curve. In a normal ventricle, the spread of the minimum volume points is narrow, leading to a low standard deviation from the mean time (see Figure 12-2). The volume curves may be used to identify regional wall motion abnormalities due to differences in amplitude between segments (Figure 12-3). A septal flash also is readily identified using the volume curves (Figure 12-4).

Figure 12-3 Time-volume curves in a patient with impaired left ventricular systolic function. A, The time-volume curves are relatively flat, with little change in absolute volume throughout the cardiac cycle, indicating severe left ventricular systolic dysfunction. B, The amplitude of the normalized volume curves is greater, making the minimum volume point easier to identify. The dispersion of the minimum volume points is greater compared with the patient with normal left ventricular systolic function. This indicates a high systolic dyssynchrony index and therefore significant dyssynchrony. The mean minimum volume is approximately 85% of the maximum, suggesting an ejection fraction in the region of 15%.

Figure 12-4 A, In this patient with ischemic left ventricular systolic dysfunction, the time-volume curves are relatively flat, although the U shape of some curves is preserved, suggesting regional wall motion abnormalities. B, When the normalized volume curves are inspected, the minimum volume of the septal and anteroseptal curves (green and light blue curves) occur at the beginning of systole. This is consistent with a septal flash.

Contraction Front Mapping

Contraction front mapping analyzes temporal and spatial activation of left ventricular contraction, representing the myocardial segments that reach peak contraction every 25 ms on a color-coded polar map of the LV. As the contraction front spreads through the ventricle, the color changes from red to blue, allowing areas of late contraction to be easily identified. The normal left ventricular activation pattern is homogeneous and rapid (Figure 12-5; Video 12-3). The activation pattern in patients with left ventricular systolic dysfunction is variable, but generally a U-shaped activation pattern is noted (Figures 12-6 and 12-7; Videos 12-4 and 12-5). In our experience, the region of maximal delay most commonly appears to be the septum, leading to a predominant left-to-right color transition. No definite pattern has been established in cases of left bundle branch block, although a septal flash may be identified in some cases (Figure 12-8; Video 12-6). This tool also has been used to identify regional wall motion abnormalities at rest and during stress echocardiography.

Figure 12-5 Normal static map showing homogeneous transition with a large area through white to blue (see Video 12-3). Ant, anterior; Int, interior; Lat, lateral; Sept, septal.

Figure 12-6 Typical static map in a dyssynchronous patient showing U-shaped earliest color transition predominantly in the basal lateral and inferolateral segments, suggesting a predominant left-to-right activation. The activation also is slower than normal; a large part of the map remains red (see Video 12-4). Ant, anterior; Int, interior; Lat, lateral; Sept, septal.

Figure 12-7 Atypical static map showing U-shaped right-to-left activation, with first color transition predominantly in the basal anteroseptal and anterior segments (see Video 12-5). Ant, anterior; Int, interior; Lat, lateral; Sept, septal.

Figure 12-8 A clear septal flash is seen (dark blue). This is again associated with a typical predominant left-to-right transition; the next activation is seen in the basal lateral and inferolateral segments (white) (see Video 12-6). Ant, anterior; Int, interior; Lat, lateral; Sept, septal.

Technique for Left Ventricle Analysis

LV analysis may be performed with a number of software packages, all following similar principles, although the results are not directly comparable due to differences in methodology and software algorithms. Described are the steps of LV analysis using 4D LV Analysis software (TomTec, Munich, Germany).

1. The end-systolic and end-diastolic frames are identified, and the axis of the LV is identified using the apex and mitral valve annulus as landmarks (Figure 12-9; Video 12-7).

2. The endocardial border is identified at end systole and end diastole in three cut frames corresponding to the traditional 2D apical views (four-chamber, two-chamber, and long-axis views). This is performed in the same manner as Simpson’s biplane view, excluding the endocardial trabeculae and papillary muscles. There is usually a tradeoff between the accuracy of estimated volumes compared with MRI and accuracy of tracking. Placing the border more subendocardially to mid-myocardially improves the correlation between 3D echocardiography and MRI volumes. However, this may be at the cost of accurate tracking, leading to a less accurate SDI. Our practice is initially to place the borders further out. However, if tracking proves inaccurate, we move the border closer to the blood-tissue interface (Figure 12-10; Videos 12-8 to 12-10).

3. These data are then used to generate a model of the LV throughout the cardiac cycle using endocardial border detection software (see Figure 12-1). The model is inspected to ensure accurate identification of the endocardial border and reliable tracking throughout the cardiac cycle. This is done in the short-axis and again in the long-axis views. This step is particularly important because accurate border tracking is essential for identification of the exact timing of the end-systolic subvolumes (Video 12-11).

4. If necessary, tracking may be adjusted in a number of ways (Figure 12-11). Initially, the contour detection sensitivity may be adjusted. This step makes the model more “stiff” because it increases dependence on the user-defined planes (Video 12-12). If tracking remains suboptimal, we return to the endocardial border identification stage and adjust the initial cut plane borders in regions where tracking is poor. It sometimes is necessary to define the border closer to the blood-tissue interface, although this does decrease the accuracy of the volume measurements (Video 12-13). Alternatively, the tracking may be adjusted on the Beutel model on a frame-by-frame basis. This method can increase the jerkiness of the Beutel model that is generated, again making the minimum volume more difficult to identify. This difficulty can be overcome by the “smoothing” function (Video 12-14).

5. Once a model that tracks satisfactorily has been achieved, ESV, EDV, EF, and SDI are calculated. At this point, we inspect the individual time-volume curves to ensure that the minimum volume has been accurately identified (Video 12-15). We find this easiest using the normalized time-volume curves because the greater amplitude of the curves makes the minimum volume clearer to identify (see Figures 12-2 to 12-4). If the minimum volume has been attributed to the wrong time, this is adjusted, changing the overall SDI. This process is repeated for both the 16-segment and 17-segment models. If there are outlier curves, the tracking in these segments should again be carefully inspected to ensure its accuracy. Segments that do not track well may be excluded from the final result.

Figure 12-9 Aligning the central axis of the left ventricle. The dataset is manipulated so the axis runs from the apex to the center of the mitral valve annulus in all three cut planes. End-systolic and end-diastolic frames are identified (see Video 12-7).

Figure 12-10 The endocardial border is traced at end systole and end diastole in three cut planes (apical four chamber, A; two chamber, B; and long axis, C). Trabeculae and papillary muscles are included in the left ventricular cavity (see Videos 12-8 to 12-10).

Figure 12-11 Methods to adjust tracking include changing the contour sensitivity level, adjusting the contour detection, and directly editing the Beutel model on a frame-by-frame basis.

Challenges in Left Ventricle Analysis

Operator Training and Experience

Assessment of left ventricular function is known for the difficulties in achieving consistent results. Formal teaching interventions have been shown to improve interobserver variability in the 2D assessment of EF, thereby ensuring good-quality results.²² Our experience supports this conclusion in the case of 3D analysis of the LV. We have found that consistent methods among operators plays an extremely important role in achieving reproducible results because even small differences in methodology may lead to a systematic bias in results, consistent with published evidence.²³ At our institution, all operators are required to undergo a formal training program using a set of 3D datasets representing a spectrum of pathologies and dataset quality. Good agreement with the results of an expert operator is required prior to beginning clinical analysis. This method has proven to be effective for us, as evidenced by the excellent interobserver and interhospital agreement in the previously mentioned study by Kapetanakis et al.⁹

Dataset Quality

As seen in the study by Soliman and colleagues,¹⁰ dataset quality has a large bearing on the accuracy and reproducibility of LV analysis.¹⁰ Mor-Avi and colleagues²³ also noted poor spatial resolution as a factor accounting for discrepancies between operators. A recent study in a “real world” population by Miller and associates²⁴ has confirmed that image quality has a significant impact on observer variability. Our experience is significantly in keeping with this finding. Up to 15% of datasets may be rejected because of poor quality, with the proportion of unusable scans higher among patients with heart disease.¹⁴

The acquisition of the raw 3D dataset should be optimized for maximal volume rate and spatial resolution. This is achieved by adjusting depth and sector settings to include only the LV, the mitral valve annulus, and the aortic valve (required for spatial orientation). Acquisition is performed during sustained breath-hold to minimize translation artifacts. Optimal datasets are achieved at heart rates less than 80 beats/min. In patients with an irregular rhythm, such as atrial fibrillation or frequent ventricular ectopy, we attempt to time acquisition to obtain volumes with a regular R-R interval. Datasets that include an ectopic beat seen on the accompanying electrocardiogram can almost universally be excluded because of the presence of stitching artifact. This process frequently requires many attempts to obtain a good-quality dataset.

In all cases, we obtain multiple datasets and inspect them for dataset quality and stitching artifact at the point of acquisition. Datasets with stitching artifact are rejected. Datasets with two or more missing segments are interpreted with caution, and a comment is made regarding dataset quality when reporting the results.

Left Ventricle Analysis in Patients with Reduced Systolic Function

LV analysis is more challenging in patients with reduced LV function. This is caused by a number of factors. First, reduced LV function is associated with LV dilation resulting in larger left ventricular volumes, which are technically more challenging to acquire (Figure 12-12; Video 12-16). The apical echocardiographic window frequently is displaced, and the apex and anterior wall become more difficult to image. Larger volumes result in lower volume rates, a problem that can be partly overcome by four- or seven-beat acquisitions. Alternatively, the high volume rate–low scan line density acquisition option may be used, although it results in reduced spatial resolution.

Figure 12-12 An example of poor image quality. Dilation of the left ventricle (LV) displaces the LV apex and makes the anterior wall more difficult to image. As the volume increases, it becomes difficult to include the entire LV in the dataset, although this can be partially overcome by increasing the number of cardiac cycles included in the acquisition. Another approach is to use the high volume rate–low scan line density option for acquisition, but spatial resolution will suffer, leading to reduced tracking accuracy (see Video 12-16).

Second, the increasing sphericity index as the LV dilates makes the true apex more difficult to identify and hence the central axis of the LV difficult to align (Figure 12-13; Video 12-17). Because the LV model is subsegmented around this axis, variation in the axis leads to alteration in subsegmental volumes and time to minimum volume.

Figure 12-13 Spherical left ventricle (LV). The central axis of the LV becomes difficult to align as the LV dilates and becomes spherical, making the true apex more difficult to identify. This may lead to inaccuracy in the minimum volumes (see Video 12-17).

Third, the reduction in LV function is associated with prolonged isovolumic times, which make the true end-systolic and end-diastolic frames more difficult to identify. Small absolute changes in ventricular volumes over the cardiac cycle contribute to this. As ventricular systolic function worsens and the difference between end-systolic and end-diastolic volumes decreases, small discrepancies in volume measurements result in a larger percentage error in EF (Video 12-18).

Fourth, the small changes in left ventricular volumes over the cardiac cycle mean that the amplitude of endocardial motion is small and therefore difficult to track. This may lead to a low signal/noise ratio, which can make the minimum volume difficult to identify. To circumvent this problem, we inspect the normalized regional volume curve rather than the lower amplitude absolute regional volume curve. We inspect each curve individually to ensure maximum curve amplitude (Video 12-19).

Finally, segments that do not track well may be reflected as noisy regional volume curves, making the minimum segmental volume impossible to identify. This problem may be partially overcome by ensuring optimal endocardial border tracking, by manual correction if necessary, and careful inspection of regional volume curves to ensure correct identification of the volume nadir. An outlier timing in this context may falsely increase SDI; consideration should therefore be given to excluding segments such as these.

Endocardial Border Detection

In some cases, endocardial border detection may prove difficult, particularly in patients with prominent papillary muscles or endocardial dropout over the cardiac cycle.

Future Directions

Development of new software such as speckle tracking potentially allows for more accurate endocardial border tracking. 3D speckle tracking software is now available from a number of vendors and allows assessment of left ventricular global and segmental volumes as well as novel quantification of measures of myocardial deformation. This uses several of the benefits of 3D LV analysis, overcoming 2D limitations such as foreshortening and geometric assumptions of the LV, and offers the added benefit of tracking of speckles moving in all three dimensions (Figure 12-14; Videos 12-20 and 12-21). Initial experience with this software suggests a quicker and simpler workflow. This approach shows promise, with good reproducibility noted in one study using software from a single vendor.²⁵ However, it requires further validation prior to widespread use.

Figure 12-14 Three-dimensional strain in a patient with impaired left ventricular systolic function. The ejection fraction is 24% and global longitudinal strain is –3.5% (see Videos 12-20 and 12-21).

Case 1: Normal Left Ventricular Synchrony with Three-Dimensional Echocardiography and Broad QRS

A 65-year-old man with longstanding dilated cardiomyopathy, attributed to a past history of viral myocarditis, presented to the CRT assessment clinic for consideration of upgrade of single-chamber defibrillator after a deterioration in heart failure symptoms. At the time of assessment, resting ECG showed sinus rhythm with left bundle branch block and QRS duration of 141 ms (Figure 12-15). However, 3DE did not suggest significant dyssynchrony, with an SDI-16 measured at 5.6% (Figure 12-16; Video 12-22). He was offered a CRT upgrade on the basis of deteriorating symptoms to New York Heart Association (NHYA) functional class III, broad QRS, and EF measured at 13%. However, the lack of 3D dyssynchrony suggested possible lack of benefit of CRT.

Figure 12-15 Electrocardiogram of the patient in Case 1, showing left bundle branch block with a QRS duration of 141 MS.

Figure 12-16 In this patient with a history of viral myocarditis, the volume curves (A) are of low amplitude. The normalized volume curves (B) show that the timing of the minimum volume is not widely dispersed, suggesting that the ventricle is not very dyssynchronous despite the electrocardiogram, which shows left bundle branch block. The static polar map (C) shows predominant right-to-left activation. These findings are reflected in the results (D), which show an intermediate systolic dyssynchrony index (SDI) of 5.6%, suggesting that this patient is unlikely to benefit from cardiac resynchronization therapy (see Video 12-22). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

One month after implantation, the patient presented again with worsening heart failure symptoms. On interrogation of his device, he was found to be 93% BiV paced. LV analysis was repeated, and SDI-16 was found to be 6.3% during BiV pacing (Figure 12-17; Video 12-23) compared with 5.2% during intrinsic rhythm (Figure 12-18; Video 12-24). The pacing function of the device was switched off. The patient was referred for ventricular assist device and was placed on the waiting list for cardiac transplantation.

Figure 12-17 After cardiac resynchronization therapy in the patient in Case 1, the volume curves remain flat (A). The dispersion of the time of the minimum volume appears to have increased slightly (B); this is reflected in the slight increase in systolic dyssynchrony index (SDI) to 6.4% (D). The activation pattern is more homogeneous but is slower (C) (see Video 12-23). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Figure 12-18 A to C, After cardiac resynchronization therapy with biventricular pacing switched off in the patient in Case 1, there is no significant change compared with biventricular pacing on. The activation pattern is once again predominantly right-to-left and occurs more rapidly. The systolic dyssynchrony index (SDI) is shown in D (see Video 12-24). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Case 2: Atrial Fibrillation

A 61-year-old man with a history of permanent atrial fibrillation, coronary artery disease, and previous coronary artery bypass grafting was referred for consideration of CRT. He had a history of increasing shortness of breath on exertion and paroxysmal nocturnal dyspnea. Adequate 3DE volumes were obtained for analysis and demonstrated severely reduced LV systolic function, with an EF of 13.5%, and gross dyssynchrony, with an SDI at 19.1% (Figure 12-19; Video 12-25). Following CRT-D implantation, he was noted to have poor ventricular rate control despite high-dose β-blocker therapy, with rates ranging from 90 to 100 beats/min. This resulted in inadequate BiV pacing therapy, at just 34% BiV pacing (Figure 12-20; Video 12-26). Atrioventricular node ablation was performed and resulted in 100% BiV pacing. LV analysis performed 4 days after his atrioventricular node ablation showed that the EF had increased, with reduction in SDI to 10.2%, suggesting significant benefit from BiV pacing. This improved further at a follow-up visit 3 months later, with improvement in EF to 30% and reduction in SDI to 5.5% (Figure 12-21; Video 12-27). Symptomatically, the patient had an excellent benefit following his two procedures.

Figure 12-19 Prior to cardiac resynchronization therapy in the patient in Case 2, adequate volumes are obtained for analysis in a patient with permanent atrial fibrillation. The volumes are carefully inspected to outrule stitching artifact prior to analysis. The volume curves (A) are flat, consistent with reduced left ventricular systolic function (ejection fraction of 13.5%). Inspection of the normalized volume curves (B) facilitates identification of the minimum volume. C, The time of the minimum volume is broadly dispersed, with the minimum volume identified at the beginning of systole in the basal to mid-septal and basal inferior segments (green and yellow curves). This represents a septal flash and is clearly seen on the static contraction map, with early septal breakthrough. D, The systolic dyssynchrony index (SDI) is high at 19.1%, suggesting probable benefit from cardiac resynchronization therapy. Note that segment 13 is excluded from the analysis in this case because of poor tracking of the endocardium (see Video 12-25). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Figure 12-20 A, Early after cardiac resynchronization therapy in the patient in Case 2, there is a small increase in the amplitude of the volume curves. However, when the normalized curves are inspected (B), the minimum volume is seen to occur early in the septal and basal inferior segments, suggesting persistent dyssynchrony. A persistent septal flash is seen on the static contraction map (C), and the systolic dyssynchrony index (SDI) remains high (D). On clinical evaluation, the patient was noted to have rapidly conducted atrial fibrillation, leading to a low percentage of biventricular pacing (see Video 12-26). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Figure 12-21 After atrioventricular node ablation, the patient in Case 2 underwent biventricular pacing 100% of the time. Repeat left ventricle analysis conducted 3 months after the procedure showed a marked increase in the amplitude of the volume curves. The segmental minimum volumes timings (A) are seen to occur closer together on the normalized volume graph (B). The early septal nadir also has disappeared. The septal flash has disappeared on the static polar map (C). Significant reverse remodeling has occurred. D, Successful resynchronization therapy is reflected in the near-normal systolic dyssynchrony index (SDI) of 5.5% (see Video 12-27). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Case 3: Suboptimal Lead Position

A 77-year-old woman with a history of idiopathic dilated cardiomyopathy, with broad QRS duration and significant dyssynchrony on 3DE assessment (SDI, 16.1%), underwent CRT-defibrillator implantation (Figure 12-22; Video 12-28). Over the subsequent 6 months, she noted a progressive deterioration in heart failure symptoms. Repeat 3DE analysis of the LV indicated persistent dyssynchrony (Figure 12-23; Video 12-29). Postimplantation chest radiograph and coronary sinus venography films from the device were reviewed, and it was noted that the LV lead had been placed in an anteroseptal branch. An alternative posterior vein was noted for lead deployment, and the patient underwent lead repositioning (Figure 12-24). Following device revision, her clinical condition considerably improved, with reduced dose of diuretic and improved exercise capacity. Subsequent repeat measurement of left ventricular SDI demonstrated a significant improvement in dyssynchrony (Figure 12-25; Video 12-30).

Figure 12-22 Prior to cardiac resynchronization therapy in the patient in Case 3, the volume curves are low amplitude (A), and there is significant dispersion of the timing of the minimum volume seen on the normalized volume curves (B), with early minimum volume of the septum (green curves). This represents a septal flash, which is seen on the static polar map (C), and there is significant dyssynchrony (D), suggesting a probable benefit from cardiac resynchronization therapy (see Video 12-28). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Figure 12-23 After cardiac resynchronization therapy in the patient in Case 3, there is no significant improvement in the volume curves (A), and the minimum volume timings remain broadly dispersed (B). The contraction pattern has changed and now is seen to spread from left to right (C), but the ventricle remains dyssynchronous, and ventricular function has not improved (D) (see Video 12-29). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Figure 12-24 A, Initial left ventricular lead position in the patient in Case 3 as viewed in left anterior oblique projection. B, Comparable view of the left ventricular lead position after procedure to resituate the lead.

Figure 12-25 Left ventricle (LV) analysis conducted 3 months after repositioning of the left ventricular lead in the patient in Case 3. The amplitude of the volume curves has increased significantly (A) and the range minimum volume time is narrower (B), demonstrating resynchronization. A homogeneous pattern of activation is seen on the static contraction map (C), and there is evidence of significant reverse remodeling (D). This correlated with an excellent symptomatic improvement (see Video 12-30). Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.

Case 4: Dyssynchrony Despite Normal QRS Duration

An 80-year-old man with a background of coronary artery disease and previous coronary artery bypass grafting was referred for consideration of implantable cardioverter-defibrillator therapy on primary prevention grounds. He had a history of exertional breathlessness and was in NYHA class III. He was known to have left ventricular systolic impairment, with an EF of approximately 30%. During his assessment, QRS duration was noted to be 96 ms, with delayed precordial R-wave progression (Figure 12-26); 24-hour ECG monitoring revealed four beats of ventricular tachycardia. The patient was referred for 3DE, which showed an EF of 33% and an SDI of 18.8% (Figure 12-27; Video 12-31). This result suggested potential benefit from CRT, although falling outside current guidelines; reimbursement therefore was subsequently denied.

Figure 12-26 The electrocardiogram in the patient in Case 4 demonstrates sinus rhythm with normal QRS duration. Despite the narrow QRS complex, three-dimensional left ventricular analysis shows significant dyssynchrony.

Figure 12-27 The regional volume curves demonstrate left ventricular systolic dysfunction, with regional wall motion abnormalities, in the patient in Case 4. The inferoposterior, septal, and anteroseptal volume curves are flat, with some preservation of the amplitude of the anterolateral (red and dark blue) curves. There is a prominent septal flash seen on the normalized volume curves (A), with the minimum volume of the septal and anteroseptal curves (green and light blue curves) detected at the beginning of systole (B). C, The static map demonstrates predominant left-to-right activation, with representation of the akinesis of the septum and anteroseptum (remains red), and septal flash seen in segment 1 only. D, There is clear dyssynchrony despite the normal QRS duration, suggesting that the patient may derive benefit from cardiac resynchronization therapy. Ant, anterior; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; Int, interior; Lat, lateral; Sept, septal; SV, systolic volume.