Diagnostic accuracy and clarity of steady-state free precession imaging of cardiac valve morphology in congenital heart disease
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/ACVI.ACVI_14_18
Keywords: Cardiac magnetic resonance imaging, cardiac valve, congenital heart disease, diagnostic accuracy
Evaluation of cardiac valve morphology has historically not been considered an indication for cardiac magnetic resonance imaging (MRI) due to suboptimal imaging quality.[1] Imaging techniques using gradient echo although illustrate valve motion and location of flow jets are limited by significant blurring and have relatively poor spatial resolution.[1] Spin echo or black-blood sequences do have better spatial resolution and can outline vessel walls, and leaflets are nonetheless “still-frame” images and have essentially no information regarding valve function.[2],[3] MRI techniques utilizing cine cardiac magnetic resonance-steady-state free precession (CMR-SSFP) may have overcome these limitations. Little evaluation has been performed of CMR-SSFP diagnostic imaging quality in the assessment of valve morphology for pediatric/congenital heart disease to the best of our knowledge. The few studies that exist have been limited by small sample size.[4],[5],[6] Other studies have primarily utilized velocity-encoded cardiovascular magnetic resonance to examine valvar pathologies such as stenosis or regurgitation or have compared agreement of two-dimensional measurements.[7],[8],[9],[10],[11],[12],[13],[14] In addition, the comparison of the diagnostic accuracy CMR-SSFP imaging with respect to aortic valve morphology to the “gold standard” of echocardiography has not been previously performed. A finding that indicates that CMR-SSFP is accurate in defining valvar morphology may expand the clinical applications of CMR-SSFP; cardiac MRI may be a viable noninvasive imaging alternative to assess cardiac valve morphology among those patients with limited transthoracic echocardiography imaging windows. In this study, we developed an imaging diagnostic clarity score and applied it to the tricuspid, mitral, and aortic valves. Our objective was to determine the diagnostic clarity of cine CMR-SSFP imaging of cardiac valves. We also examined the accuracy of cine CMR-SSFP in correctly diagnosing aortic valve morphology. In addition, we attempted to identify preferable imaging planes for assessment of various valve components. We hypothesized that cine CMR-SSFP images would yield high-quality diagnostic images in the majority of cases and would have a high degree of diagnostic accuracy of aortic valve morphology.
Data source We retrospectively reviewed 234 consecutive pediatric/congenital cardiac MRI studies performed at Boston Children's Hospital. Imaging protocols All studies were performed using a 1.5T clinical MRI scanner (Philips Achieva; Philips Medical Systems, Best, The Netherlands). Surface coils were selected based on patient size. Cardiac gating and heart rate monitoring were performed with vector electrocardiography (ECG) gating. The CMR examination protocol included breath-hold cine imaging using an ECG-gated SSFP pulse sequence in two-chamber, four-chamber, and short-axis planes for quantification of biventricular size and ejection fraction. Aortic valve imaging was performed in an oblique coronal [Figure 1]a, oblique sagittal [Figure 1]b, and short axis to the aortic root [Figure 1]c following our standard laboratory protocol. Techniques to achieve these imaging planes have been previously described.[1]
Morphologic analysis of valves For diagnostic clarity analysis, the atrioventricular valves (mitral and tricuspid valves) were divided into four anatomic components: (1) annulus, (2) leaflets, (3) chordae tendineae, and (4) papillary muscles. Aortic valves were divided into two components: (1) annulus and (2) leaflets/commissures. We assessed the diagnostic clarity of the mitral and tricuspid valves by examining their valve components using standard cine CMR-SSFP imaging in two-chamber, four-chamber, and short-axis views. There were no additional specific imaging angles performed to enhance the illustration of these valve components. We assessed diagnostic clarity of the aortic valve when long-axis (oblique coronal and oblique sagittal) and short-axis imaging of the aortic valve and root was performed. Diagnostic clarity score We assigned a diagnostic clarity score 1 through 5 [Table 1] to the each of the valve components (annulus, leaflets, chordae tendineae, and papillary muscles) of the tricuspid and mitral valves and assigned a diagnostic clarity score to the annulus and leaflets of the aortic valve. A score of 1 indicated excellent clarity and the lowest score 5 indicated nondiagnostic imaging quality. We defined high-quality diagnostic imaging as a clarity score of 1 or 2.
Interobserver diagnostic clarity score analysis We assessed the interobserver agreement of diagnostic clarity scores. A second investigator (AP) reviewed randomly selected 10% sample of studies using the same scoring system described above for interobserver agreement of clarity score assignment. The level of agreement was calculated a percentage of agreement for each valvular anatomic segment between the two reviewers (AP and OJB); both reviewers had >3 years of CMRI clinical experience. Diagnostic accuracy of aortic valve morphologic diagnosis Among patients with aortic valve imaging, a morphologic diagnosis was made utilizing only the CMR-SSFP images. The reviewers were blinded to the patients underlying diagnosis and morphologic diagnosis of the aortic valve by echocardiography. This morphologic diagnosis was then compared to the existing morphologic diagnosis established by the gold standard of echocardiography and confirmed by a blinded review of the echocardiograms. Data collected Cardiac diagnosis, patient age, gender, weight, body surface area, anatomic diagnosis, repetition time (TR), echo time (TE), and slice thickness were collected. Statistical analysis Continuous variables were expressed as median and range or mean and standard deviation, as appropriate. Anatomic diagnoses were tabulated according to the frequency. The proportion of valve components with a high-quality diagnostic imaging (diagnostic clarity scores 1 or 2) was expressed as a percentage for the tricuspid, mitral, and aortic valves. We compared patient age at the time of the scan, gender, weight, body surface area, anatomic diagnosis, TR, TE, and slice thickness between those cases with a high diagnostic clarity score (score 1 or 2) versus those with lower diagnostic clarity scores (score 3, 4 or 5) using the Wilcoxon rank-sum test. Interobserver agreement was expressed as a percentage of agreement on the assignment of a high diagnostic clarity score for all tricuspid, mitral, and aortic valve components. An average composite score of all valve components was calculated for each valve. For example, each valve component could have a score of 1 through 5 assigned; therefore, the best average score a valve could have would be a 1 and the worst would be a 5. The average composite scores for the tricuspid, mitral, and aortic valves were plotted on a histogram according to their frequency and average score. This was done to illustrate the distribution of overall imaging clarity each valve. The proportion of correct aortic valve morphologic diagnoses made by CMR-SSFP using echocardiography as a gold standard was expressed as a percentage of agreement.
Patient and study characteristics There were a total of 233 congenital/pediatric cardiac MRI studies evaluated with a total of 1892 valve components. [Table 2] illustrates the 20 most common diagnoses during the study period. Together, these 20 diagnoses represent 89% of all patients in the study. Studies of patients with tetralogy of Fallot, coarctation of the aorta, and connective tissue disorders represented one-third of all cases. [Table 3] illustrates patient and study characteristics. Patient age ranged from 1 month to 65.1 years (average: 17.2 years). Females represented 59% of the sample and average patient weight was 58.1 kg with a range of 2.4-99.8 Kg. The median slice thickness for four-chamber and short-axis imaging was 8 mm (range: 5-8 mm). The median slice thickness for aortic valve imaging was 7 mm (range: 5-8 mm). The overall TE range was 1.28-1.83 ms. Diagnoses present in the sample are also listed.
Diagnostic clarity scores of valve components [Table 4] illustrates the proportion of valve components with a high diagnostic clarity score. We found that the majority of valve annuli, leaflets, and papillary muscles had high diagnostic clarity. Most of these components had a high clarity score 64%-80% of the time - the tricuspid valve papillary muscles had a high diagnostic clarity score 53% of the time. The valve components that were least likely to have a high diagnostic clarity score were the chordae tendineae of the mitral and tricuspid valves.
We examined patient and scan parameters to look for patient and scan characteristics associated with high diagnostic clarity score. We found that there was no difference in weight or body surface area, slice thickness, or TE between those with an average mitral or tricuspid clarity score of <2 (excellent or very good clarity) versus ≥2 (fair to poor clarity). Interobserver agreement of diagnostic clarity scores [Table 5] illustrates the percentage of interobserver agreement of diagnostic clarity scores. The aortic valve had excellent interobserver agreement - 85% agreement for the aortic valve annulus and leaflets. The mitral valve components also demonstrated very good agreement: annulus (86% agreement), leaflets (68% agreement), and papillary muscles (81% agreement). The tricuspid valve annulus and leaflets also had good agreement (77% and 59%). The chordae tendineae of the mitral and tricuspid valves and the tricuspid valve papillary muscle had the lowest interobserver agreement.
Average composite valvar diagnostic clarity score For each case, we calculated the average of diagnostic clarity scores of all components of the tricuspid, mitral, and aortic valves. [Figure 2]a, [Figure 2]b, [Figure 2]c illustrates the distribution of this composite average quality scores for the tricuspid, mitral, and aortic valves, respectively. For tricuspid valves, the average score was ≤2 (high imaging quality), 40% of the time, and had an average composite score ≤3 (indicating high quality and good quality/diagnostic images), 80% of the time. For mitral valves, the average score was ≤2 (high imaging quality), 50% of the time, and ≤3, 82% of the time. For aortic valves, the average score was ≤2 (high imaging quality), 63% of the time, and had an average composite score ≤3, 92% of the time.
Diagnostic accuracy of aortic valve morphology Among the 39 cases with aortic valve-specific imaging, 27 (69%) had an echocardiogram for comparison. CMR-SSFP correctly identified the aortic valve morphology in all 27 cases. CMR-SSFP correctly identified 14 of 14 tricommissural aortic valves [Figure 3]a and correctly identified the affected commissure in 13 of 13 bicommissural aortic valves [Figure 3]b and [Figure 3]c.
Little has been written on the diagnostic clarity and diagnostic accuracy of CMR-SSFP imaging-related cardiac valve morphology. There have been prior studies comparing two-dimensional measurements of CMR to echocardiography.[15],[16],[17] However, to the best of our knowledge, this study is the first to develop and apply a diagnostic clarity score for multiple cardiac valve components and to compare the diagnostic accuracy of cine CMR-SSFP imaging to echocardiography as a gold standard. Diagnostic clarity scores of valve and valve components In the majority of cases, CMR-SSFP produced high diagnostic quality imaging of cardiac valve morphology in congenital/pediatric cardiac MRI. The valve components with the highest diagnostic clarity score are tricuspid and mitral valve annuli, leaflets and papillary muscles, and aortic valve annuli and leaflets. The average composite diagnostic clarity score also indicated that both high-quality imaging and good imaging of diagnostic quality can be achieved in the majority of cases. Having a diagnostic clarity with cine CMR-SSFP may help in identifying valvar pathology such as aortic valve defects, among patients with limited echocardiography imaging windows. In such cases, cine CMR-SSFP may be used as an alternative imaging modality for surgical planning. Somewhat surprising is the high proportion of papillary muscles that have a high clarity score. This is important as papillary muscle morphology and attachments may be important in assessing abnormalities of the atrioventricular valves and may have important interventional implications. CMR-SSFP, however, produced high-clarity images of chordae in only a minority of cases. These findings may not be surprising as the spatial resolution, and signal needs are great to the chordae tendineae, which are thin structures. The lack of significance between small patients and larger patients with respect to achieving a high diagnostic clarity score suggests that high-quality valve images may be possible in smaller patients. Interobserver agreement of diagnostic clarity scores The interobserver agreement was greatest for those structures with higher clarity scores (valve annuli and leaflets and mitral valve papillary muscles) and lowest for the valve structures with lower clarity scores such as the chordae tendineae and the papillary muscles of the tricuspid valve. These findings are intuitive as it may be easier for observers to achieve good agreement on structures that are well visualized. This finding suggests that diagnostic findings related to the valve annuli and leaflets and of the mitral valve papillary muscles may be fairly reliable and reproducible among MRI cardiac images in the clinical setting. Diagnostic accuracy of aortic valve morphology Aortic valve morphology can be diagnosed with a high degree of reliability. Not only was the diagnosis of bicommissural aortic valve correctly made in all cases but also the affected commissure was accurately identified with excellent agreement with echocardiography. These findings suggest that cine CMR-SSFP may be used to assess aortic valve morphology among patients with poor echocardiography windows and may be the preferred noninvasive imaging alternative. Imaging planes for assessment of valve components We noted that, as in echocardiography, certain views are better suited for assessment of different valve components. In general, no single imaging plane is sufficient to assess valve morphology. The four-chamber view provides a very good imaging plane for assessment of the atrioventricular valve annuli. The two-chamber view was occasionally off-center, thereby limiting the view of the annuli. The leaflets were seen well in multiple views; however, leaflet motion was best seen in the two-chamber (when centered properly) and four-chamber views. The short-axis images were good in identifying leaflet morphology. The chordae and papillary muscles were best seen in the short-axis views. The short-axis plane was allowed for easier assessment of the course and insertion points of the chordae and papillary muscles than in the long-axis views of the heart where through-plane motion of these structures limited their visualization. Advantages and disadvantages of cardiac magnetic resonance-steady-state free precession valve imaging An advantage of CMR-SSFP sequence is good signal-to-noise ratio and valve structure-to-blood pool contrast. Acquisition times are also somewhat shorter and high-quality imaging can be acquired using breath-hold imaging. One disadvantage CMR-SSFP imaging may have relative to gradient-echo cine sequences is visualizing flow jets (regurgitant jets primarily), which result from dephasing of turbulent flowing blood.[1] Limitations This study may be limited in its retrospective nature in that only standard imaging planes were used to examine valve morphology. We did not explore if additional or other imaging planes or variation in scan parameters (such as signal-to-noise ratio, spatial or temporal resolution, or degree of anatomic coverage) would have yielded better diagnostic clarity. An additional limitation is that these images were performed with breath-hold imaging, and therefore, these findings might not be generalizable to cine CMR-SSFP studies performed as breathe-through cine images.
Cine CMR-SSFP imaging yields high quality, reproducible, diagnostic imaging of the tricuspid, mitral, and aortic valves and has high diagnostic accuracy for aortic valve morphology. Cine CMR-SSFP imaging is a viable and in some cases preferable alternative to noninvasive imaging of cardiac valve morphology. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]
|
| Similar in PUBMED |
| Search Pubmed for |
Search in Google Scholar for
|
| Related articles |
| Access Statistics |
| Email Alert * |
| Add to My List * |
| * Registration required (free) |
| In this article |
| Abstract |
| Introduction |
| Materials and Me... |
| Results |
| Discussion |
| Conclusions |
| References |
| Article Figures |
| Article Tables |
| Article Access Statistics | |
| Viewed | 4022 |
| Printed | 252 |
| Emailed | 0 |
| PDF Downloaded | 403 |
| Comments | [Add] |
