Analysis of left ventricular mass, volume indices and ejection fraction from. SSFP cine imaging: a comparison of semi-automated, simplified manual,

Transcription

1 Analysis of left ventricular mass, volume indices and ejection fraction from SSFP cine imaging: a comparison of semi-automated, simplified manual, detailed manual and geometric modelling techniques Christopher A Miller 1,2,4 *, Peter Jordan 1, Alex Borg 1, Rachel Argyle 1, David Clark 3, Keith Pearce 1 and Matthias Schmitt 1,4 1. Division of Cardiology and Cardiothoracic Surgery, University Hospital of South Manchester, Wythenshawe, Manchester M23 9LT, UK 2. Cardiovascular Research Group, The University of Manchester, Oxford Road, Manchester M13 9PL, UK 3. Alliance Medical, Wythenshawe CMR Unit, Wythenshawe, Manchester M23 9LT, UK 4. Biomedical Imaging Institute, The University of Manchester, Oxford Road, Manchester M13 9PL, UK *Corresponding author addresses: CAM: chrismiller@doctors.org.uk PJ: pjordan@doctors.org.uk AB: dr.alex.borg@googl .com RA: Rachel.Argyle@pat.nhs.uk DC: daclark@alliance.co.uk KP: keith.pearce@uhsm.nhs.uk MS: matthias.schmitt@uhsm.nhs.uk 1

2 Abstract Background: Cardiovascular magnetic resonance (CMR) is considered the gold standard for assessment of left ventricular (LV) mass, volume indices and ejection fraction (EF) however variability exists due to LV base positioning, end- systolic frame identification and method of endocardial contouring. The aims of this study were first to compare 3 commonly used analysis techniques; assessing mass, volume indices and EF obtained with each and their reproducibility, and second to assess the performance of 6 LV geometric models. Methods: Steady- state free precession images from 50 consecutive patients were analysed with 3 techniques; 1. Semi- automated; including semi- automated LV base identification, end- systolic frame selection and endocardial contouring; 2. Simplified manual; including manual basal LV slice identification, end- systolic frame selection and simplified endocardial contouring (papillary muscles and trabeculae included in volumetric measurements); 3. Detailed manual; identical to Simplified manual except with manual detailed endocardial contouring (papillary muscles and trabeculae included in mass); and 6 geometric models; Teicholz, modified Simpson s, hemi- ellipse, monoplane, biplane and triplane. Bland- Altman analysis and Wilcoxon rank testing were used to evaluate the level of agreement between each method and the significance of mean differences respectively. Interobserver and intraobserver reproducibility were assessed by re- analysis of 25 (50%) scans. Results: The 3 analysis methods were not interchangeable. Simplified manual analysis significantly overestimated volumes and underestimated EF (EF underestimated by 6+4% compared to detailed manual analysis, p<0.0005). Both manual techniques significantly underestimated mass compared to semi- automated analysis (simplified manual analysis 43+20g; detailed manual analysis 34+19g; p< for both). Semi- automated LV base 2

3 position and end- systolic frame selection were significantly different from manual techniques. Semi- automated analysis showed significantly higher reproducibility than both manual techniques for measurement of EF and mass. Geometric models were not interchangeable with conventional analysis and their reproducibility was low. Conclusions: Methods for measuring LV mass, volume and EF are not interchangeable and normal reference ranges appropriate to analysis technique must be used. A technique that includes semi- automated endocardial contouring, LV base identification and end- systolic frame selection is the most reproducible. Analysis of CMR images with geometric models should be discouraged. 3

4 Background Accurate and reproducible assessment of left ventricular (LV) mass, volume indices and ejection fraction is an important strength of cardiovascular magnetic resonance imaging (CMR). These parameters remain some of the most evidenced- based indictors of prognosis, and both their absolute measurement and temporal change are used to guide pharmacological, device and surgical intervention [1-5]. Although CMR is the gold standard technique for assessment of LV mass, volume and EF, a number of factors lead to variability in image analysis. First, many institutions use only short- axis images for analysis despite it often being difficult to determine the basal LV slice, and how much of it to include, from short- axis images alone. Arbitrary criteria are widely used, such as inclusion of a slice within the LV when at least 50% or 75% of the cavity is surrounded by myocardium (Figure 1) [6-8]. Alternatively if a slice is thought to include both ventricular and atrial myocardium, others advocate tracing up to the apparent junction of atrium and ventricle before joining up the contours with a straight or curved line through the blood pool, creating a semi- circle or crescent- shaped cavity [9]. Second, it is standard practice in many centres to perform simplified endocardial contouring whereby only the compacted endocardial border is traced (Figure 2) [10, 11]. As a result, papillary muscles and trabeculae are included in cavity volumes rather than within mass. Third, end- systolic frame selection is usually performed manually by visually identifying the frame with the smallest cavity [6]. However often this is not uniform across short- axis slices, particularly in dyssynchronous ventricles (Figure 3). This study had two objectives. First, we sought to compare mass, volume indices and EF obtained using 3 commonly used analysis techniques, and the reproducibility of each 4

5 method, in a prospective cohort reflective of real- world CMR practice. The 3 techniques assessed were: 1. Semi- automated analysis; including semi- automated LV base identification, end- systolic frame selection and endocardial contouring; 2. Simplified manual analysis; including manual basal LV slice identification, end- systolic frame selection and simplified endocardial contouring; 3. Detailed manual analysis; identical to simplified manual analysis except with manual detailed endocardial contouring (Figures 1-3). We hypothesised that the 3 techniques would yield different LV mass, volume and EF values and that the techniques would differ in their reproducibility. Second, we sought to compare volumes and EF obtained with 6 LV geometric modelling techniques (Figure 4, 5), and the reproducibility of each model, with the most reproducible of the 3 methods described above. One of the advantages of CMR is that it can image the LV in its entirety so that no geometric assumptions need to be made. Nevertheless, a number of LV geometric approximation models have been validated, are widely used by other imaging modalities, are advocated for analysing CMR images and are incorporated into state- of- the- art CMR analysis software packages [8, 12-16]. We hypothesised that geometric models would yield different LV volume and EF values to conventional analysis, and would have inferior reproducibility. 5

6 Methods Patients The study was conducted according to the Helsinki Declaration. Ethical approval for the study was given by an ethics committee of the UK National Research Ethics Service (reference number 08/H1004/153) and written informed consent was obtained from all participants before entering the study. Fifty consecutive consenting patients undergoing clinically- indicated CMR scanning at a single institution (University Hospital of South Manchester) were enrolled. Patients were recruited prospectively, prior to undergoing imaging. The only exclusion criterion was known complex congenital heart disease. In order to reflect real- world clinical practice, patients were not excluded on the basis of arrhythmia or breath- holding ability. CMR imaging All patients underwent CMR imaging using a 1.5 Tesla scanner (Avanto; Siemens Medical Imaging, Erlangen, Germany) with a 32- element phased- array coil. Steady- state free precession end- expiratory breath- hold cines were acquired in 3 long- axis planes (horizontal long- axis, horizontal short- axis and 3- chamber long- axis). Subsequently, contiguous short- axis cines were acquired from the atrioventricular ring to the apex (SA stack). Typical parameters included repetition time 2.9ms, echo time 1.2ms, flip angle 80 degrees, matrix 256 x 208, in plane pixel size 1.4 x 1.4mm, slice thickness 8mm (inter- slice gap 2mm for the SA stack), temporal resolution 30-50ms, depending on heart rate. Retrospective gating was used in 48 patients. Prospective gating was used in 2 patients due to arrhythmia. Image analysis 6

7 LV mass and volumetric analysis was performed on each scan using the 3 methods described below. Analysis was carried out by a Level 2 accredited operator with over 3- years experience of clinical CMR and of each analysis method. Semi- automated analysis (Auto) Semi- automated analysis was performed using CMRtools (Cardiovascular Imaging Solutions, London, UK). The epicardial border was manually traced at end- diastole in successive short- axis slices. A second contour was placed within the myocardium on the short- axis slices at end- diastole and systole allowing the signal intensity between the two contours (i.e. the signal intensity of myocardium) to be automatically determined. Detailed, semi- automated tracing of the endocardium at end- diastole and end- systole was then performed using a signal intensity thresholding tool, such that papillary muscles and trabeculae were included in mass and excluded from volumetric measurements (Figure 2). The mitral and aortic valve positions at end- diastole and systole were manually identified on the 3 long- axis images, allowing the valve planes to be tracked through the cardiac cycle. This was integrated into the SA stack analysis allowing automated identification of the LV base and outflow tract (Figure 1). Finally, end- systole was automatically determined as being the frame with the smallest cavity volume, by calculating cavity volume at each frame (Figure 3). Simplified manual analysis (SimpMan) Simplified manual analysis was performed using Argus Syngo MR software (Siemens Medical Imaging, Erlangen, Germany). The epicardial border was manually traced at end- diastole in successive short- axis slices. The compacted endocardial border was traced on the short- axis slices at end- diastole and end- systole such that papillary muscles and trabeculae were included in volumetric and excluded from mass measurements (Figure 2). At the base of the heart, slices were considered to be within the LV at end- diastole and end- systole if the cavity 7

8 was surrounded by 50% or more of ventricular myocardium. If the basal slice contained both ventricular and atrial myocardium, contours were drawn up to the junction and joined by a curved line through the blood pool (Figure 1). The end- systolic frame was manually identified as the frame in which the LV cavity visually appeared smallest. In cases where there was a discrepancy between slices, the frame in which the more basal slices had the smallest cavity was used as it was felt this would more closely represent the smallest cavity volume. Detailed manual analysis (DetMan) Detailed manual analysis was identical to simplified manual analysis except that detailed manual tracing of the endocardium was performed such that papillary muscles and trabeculae were included in mass and excluded from volumetric measurements (Figure 2). The data from each technique was used to calculate end- diastolic (EDV) and end- systolic (ESV) volumes, from which stroke volume (SV) and EF were derived. End- diastolic myocardial mass was determined by multiplying myocardial tissue volume at end- diastole by 1.05 g/cm 3 (specific density of myocardium). Short axis slices were numbered with slice 1 being the most basal, and the most basal slice included in the analysis at end- diastole and end- systole was recorded for each technique, no matter how little of the slice was included. End- systolic frame used and the time taken for analysis was also recorded. Geometric modelling LV volumetric analysis was performed on each scan using 6 geometric models (Figure 4, 5). Appropriate contours were drawn for each model using open- source OsiriX Imaging Software. EDV, ESV and EF were calculated using an OsiriX software plug- in, with the exception of the triplane model where contours were drawn using OsiriX but calculations 8

9 were performed manually as no appropriate plug- in was available. Papillary muscles and trabeculae were included in volumetric measurements. Results obtained from each model were compared with the method found to be most reproducible of the 3 described above. LV internal diameter, defined as the distance between the endocardial border of the septum and the inferolateral wall just distal to the mitral valve leaflet tips in the 3- chamber long- axis view, was measured in diastole and systole and compared to EDV and ESV respectively. Reproducibility To assess interobserver reproducibility, 25 randomly selected scans (50%) were independently re- analysed by a second experienced observer using the methods described. To assess intraobserver reproducibility, 25 scans (50%) were re- analysed by the first observer with a 1- month temporal separation between analyses. Statistics Values are expressed as mean ± standard deviation (SD) unless stated otherwise. Histogram plots showed the data to be non- normally distributed. Agreement between each method was evaluated using Bland- Altman testing by calculating mean difference (bias) and 95% limits of agreement (i.e. mean difference + 2 SD). The significance of the differences between each method was assessed using Wilcoxon rank testing. Correlation was assessed using Spearman s rank correlation coefficient (whilst it is recognised that assessment of agreement is more appropriate than correlation, both are displayed in keeping with convention). Reproducibility of each method was assessed using the repeatability co- efficient (defined as 1.96 x [sum of the squares of the differences between observer measurements divided by n]) [17]. The significance of the differences in reproducibility was assessed using a Wilcoxon rank comparison of the squared differences [18]. P- values <

10 were considered significant. Statistical analysis was performed using SPSS Statistics (version 19, IBM). 10

11 Results Patient characteristics and scan indications were representative of those referred for CMR at our institution (Table 1). Patients demonstrated a variety of cardiovascular pathology and a wide range of mass, EDV, ESV and EF (Table 2). Comparison of semi- automated, detailed manual and simplified manual analysis Volumetric, EF and mass measurement Compared to the other two techniques, simplified manual analysis significantly overestimated both EDV and ESV, but ESV to a greater degree (Table 2 and 3, Figure 6). As a result, this method significantly underestimated SV and EF (mean difference in EF between semi- automated and simplified manual analysis 7+5%, and between detailed manual and simplified manual analysis 6+4%; p< for both). As a consequence, 6 patients (12%) classified as having a normal EF by the semi- automated technique (using a nominal normal EF cut- off of > 55%) were classified as having a reduced EF by the simplified manual technique. Furthermore, 6 patients (12%) classified as having mild to moderate LV impairment by the semi- automated technique (EF < 55% but > 35%) were classified as having severe LV impairment (EF < 35%) by the simplified manual technique. There was no significant difference between semi- automated and detailed manual analysis for EDV, ESV, SV and EF, but the 95% limits of agreement were wide for each parameter suggesting these methods were not interchangeable (Table 2 and 3). Whilst the simplified manual technique significantly underestimated mass compared with the detailed manual technique, both underestimated mass compared with the semi- automated method (mean difference between semi- automated and simplified manual 11

12 analysis was 43+20g; between semi- automated and detailed manual analysis was 34+19g; p< for both). Basal slice selection, end- systolic frame selection and analysis time Semi- automated analysis determined the base of the LV to be more basal than manual analysis at both end- diastole and end- systole. The most basal short- axis slice included in the semi- automated and manual analyses at end- diastole was and respectively (where short- axis slice number 1 is the most basal) and at end- systole was and respectively (p< for both). The frame used as end- systole was also significantly different between the semi- automated and manual techniques (frame number versus respectively; p=0.003). However, in the 21 patients where there was a difference in end- systolic frame selection, mean difference in ESV was only 2+1mls. Time taken for the simplified manual (7.0+1 minutes) and semi- automated (7.5+1 minutes) analyses were both significantly shorter than for detailed manual analysis ( minutes; p< for both). There was no significant difference between simplified manual and semi- automated analysis time. Reproducibility (Figure 7) Semi- automated analysis had higher interobserver and intraobserver reproducibility than both manual techniques for measurement of EF (interobserver; p< compared to simplified manual technique and p=0.001 compared to detailed manual technique; intraobserver; p=0.04 compared to both manual techniques) but there was no difference in reproducibility between the two manual techniques for measurement of EF. Both the semi- 12

13 automated and detailed manual techniques were significantly more reproducible than simplified manual analysis for measurement of EDV and ESV. However there was no difference in reproducibility between the semi- automated and detailed manual techniques for EDV and ESV measurements. Semi- automated analysis had higher interobserver reproducibility than both manual techniques for measurement of mass (p=0.02 compared to simplified manual technique and p< compared to detailed manual technique) although intraobserver reproducibility was not significantly different. Comparison of geometric modelling with semi- automated analysis Geometric models were compared with the semi- automated analysis technique. EDV, ESV and EF varied substantially between geometric models (Table 4). The biplane model was the most accurate compared with the semi- automated technique, with no significant difference in EDV, ESV and EF, although limits of agreement were wide for each parameter (Table 5). For the majority of the other models, volume and EF measurements were significantly different from the semi- automated technique, with very wide limits of agreement. There was a moderate correlation between LV internal diameter in diastole and EDV (R S 0.65) and a good correlation between LV internal diameter in systole and ESV (R S 0.83). Interobserver and intraobserver reproducibility for measurement of EF was low for all models and was significantly lower than that of the semi- automated technique (Figure 8). Reproducibility of EDV and ESV measurements was variable. Interobserver reproducibility for measurement of EDV using the Teicholz, triplane, modified Simpson s and biplane models was not significantly different to that of the semi- automated technique. However all models had significantly lower interobserver reproducibility than the semi- automated technique for measurement of ESV. All models also had statistically lower intraobserver 13

14 reproducibility for measurement of EDV and ESV than the semi- automated technique. 14

15 Discussion This study compared 3 commonly used techniques for measuring LV mass, volume and EF from CMR SSFP cine images. We found that analysis methods were not interchangeable and the reproducibility of the methods differed significantly. We also assessed 6 LV geometric models and found that the majority yielded significantly different volume and EF values compared to conventional analysis and all had statistically inferior reproducibility for measurement of EF. Recruitment was prospective and consecutive, without exclusion on grounds of arrhythmia or breath- holding ability and prospective gating was required in 4%. Patients exhibited a wide range of EF, volumes and mass and a variety of pathology. It is therefore felt that the study is representative of clinical CMR practice. Whilst there were no mean differences between the semi- automated and detailed manual techniques for measurement of EDV, ESV and EF, limits of agreement were wide suggesting they were not interchangeable. As expected, simplified endocardial contouring, where papillary muscles and trabeculae are included within volumes and excluded from mass, led to an overestimation of volumes compared to detailed contouring. However in keeping with other studies, ESV was overestimated to a greater degree (EDV overestimated by 5%, ESV by 19%) [19, 20]. Consequently simplified contouring led to significant underestimation of EF. The degree of EF underestimation in this study (6% difference between detailed and simplified manual contouring) is larger than that found by Weinsaft et al [20] (3%) although the current study includes patients with a wider range of volumes and EF. In a study assessing the impact of simplified versus detailed tracing of trabeculae alone, Papavassiliu et al [19] found the simplified technique underestimated EF by 2% (papillary muscles were 15

16 included within mass measurements in all subjects). The greater underestimation of ESV relative to EDV by the simplified technique may result from it being more difficult to differentiate the compacted endocardial border from papillary muscles and trabeculae at end- systole. The technique of simplified endocardial contouring is widely practiced [10, 11], despite detailed contouring (manual or semi- automated) having been used to establish normal reference ranges for SSFP imaging [6, 9, 21] and the greater accuracy of detailed contouring in ex- vivo validatory studies [22, 23]. In this study the use of simplified contouring had potential clinical implications, with nearly a quarter of patients being assigned to a different EF category compared to detailed contouring (i.e. abnormal EF rather than normal, or severely impaired EF rather than mild- moderately impaired), thus highlighting the importance of using reference ranges appropriate to the analysis method used. It appears that the method of endocardial contouring has a greater impact on volumes and EF than manual versus semi- automated LV base positioning and end- systolic frame selection. Detailed manual and semi- automated techniques had comparable endocardial contouring but significant differences in end- diastolic and end- systolic LV base positioning and end- systolic frame identification. However despite these discrepancies, there were no differences in volume and EF measurements. Conversely, there was no difference in LV base positioning and end- systolic frame identification between the simplified and detailed manual endocardial contouring techniques, but volume and EF measurements were significantly different. Simplified manual endocardial contouring led to an underestimation of mass compared to detailed manual contouring, although to a lesser degree than in the study by Weinsaft et al 16

17 [20]. The detailed manual technique also significantly underestimated mass compared with the semi- automated method, with a mean difference of 34g. This large difference appears to be in accordance with the difference seen between studies by Maceira et al [21] and Hudsmith at al [6] that defined normal values for LV mass (and volumes and EF) using different analysis techniques. Using the semi- automated method, Maceira et al found mean myocardial mass to be 156g in healthy males and 128g in healthy females but using the detailed manual technique, Hudsmith et al found mean mass to be 123g and 96g in healthy males and females respectively. The reason for this substantial difference between techniques is not clear, particularly because there were no discernable differences in epicardial contouring in the current study. It is possible however that the difference in LV base positioning has a large effect on mass measurements. This finding further highlights the importance of using normal reference ranges appropriate to analysis method used. The semi- automated technique was the most reproducible for measurement of EF and mass. With automatically determined end- systole, straightforward identification of the mitral valve plane from long- axis images (and hence LV base) and semi- automatic endocardial border tracing, the main observer variability for the semi- automated technique stems from the manual epicardial contouring (which only affects mass) and the level of the thresholding. In comparison, manual analysis requires much greater manual input (including manual endocardial and epicardial tracing, manual identification of the LV base from short axis images alone and manual identification of end- systole), explaining the inferior reproducibility. These significant differences in reproducibility between techniques are important for patient follow- up and for research study sample sizes [24]. Detailed manual endocardial contouring was significantly more reproducible than simplified manual contouring for measuring EDV and ESV, possibly because the compacted 17

18 endocardial border was more difficult to identify, however there was no difference in EF reproducibility. Papavassiliu et al [19] found no significant differences in EDV, ESV or EF reproducibility although only detailed versus simplified tracing of trabeculae was assessed (papillary muscles were included within mass measurements in all subjects). Compared with other parameters, measurement of mass showed greatest variability for all techniques, in keeping with other studies [24]. Geometric models are frequently used in echocardiography and have been advocated as a time- saving alternate to conventional CMR image analysis [16]. Mean difference and levels of agreement between each model and conventional analysis in this study are similar to the findings of Thiele et al [12] although somewhat surprisingly, the biplane rather than the triplane model showed highest EF agreement here. Nevertheless, limits of agreement were wide for all models suggesting modelling and conventional analysis were not interchangeable. Furthermore, reproducibility of EF for all models was too low for clinical or research utility, with repeatability co- efficients of 13% or greater. It is suggested that geometric models devalue CMR and their use should be discouraged. Study limitations The main limitation of this study is that no comparison is made to an externally validated gold standard, so accuracy of analysis techniques cannot be assessed. Nevertheless, the study aimed to compare mass, volume and EF values obtained by each technique and their reproducibility, which has been possible. In addition, interobserver and intraobserver reproducibility have been assessed, but not inter- study reproducibility. It is likely that inter- study reproducibility would be lower than the reproducibility demonstrated here but we hypothesis that the semi- automated technique would remain the most reproducible. 18

19 Conclusions Despite advances in all cardiac imaging modalities, the accuracy and reproducibility of CMR for assessing LV mass, volume indices and EF remains unrivalled. However this status will only be upheld if image analysis is performed to a high standard and in a consistent manner. Methods for measuring these parameters are not interchangeable and normal reference ranges appropriate to analysis technique must be used. A technique that includes semi- automated endocardial contouring, LV base identification and end- systolic frame selection was found to be the most reproducible and these features should be part of all analysis software. Use of geometric models for analysing CMR images should be discouraged. Competing interests The authors declare that they have no competing interests. Authors' contributions MS and CAM conceived the study. CAM, MS, KP and DC helped plan and design the study. CAM, PJ, AB, RA, DC, and KP collected the data. CAM analysed the data. CAM wrote the preliminary draft of the manuscript and all authors supplied comments and corrections. All authors read and approved the final manuscript. CAM and MS are the guarantors. 19

20 Acknowledgements Dr Miller was supported by a Doctoral Research Fellowship from the National Institute for Health Research, UK. Dr Schmitt was supported by Greater Manchester Comprehensive Local Research Network funding. 20

21 References 1. Ghali JK, Liao Y, Simmons B, Castaner A, Cao G, Cooper RS: The prognostic role of left ventricular hypertrophy in patients with or without coronary artery disease. Ann Intern Med 1992, 117: White HD, Norris RM, Brown MA, Brandt PW, Whitlock RM, Wild CJ: Left ventricular end- systolic volume as the major determinant of survival after recovery from myocardial infarction. Circulation 1987, 76: Pfeffer MA, Braunwald E, Moye LA, Basta L, Brown EJ, Jr., Cuddy TE, Davis BR, Geltman EM, Goldman S, Flaker GC, et al.: Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction. Results of the survival and ventricular enlargement trial. The SAVE Investigators. N Engl J Med 1992, 327: Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, Domanski M, Troutman C, Anderson J, Johnson G, et al: Amiodarone or an implantable cardioverter- defibrillator for congestive heart failure. N Engl J Med 2005, 352: Hwang MH, Hammermeister KE, Oprian C, Henderson W, Bousvaros G, Wong M, Miller DC, Folland E, Sethi G: Preoperative identification of patients likely to have left ventricular dysfunction after aortic valve replacement. Participants in the Veterans Administration Cooperative Study on Valvular Heart Disease. Circulation 1989, 80:I Hudsmith LE, Petersen SE, Francis JM, Robson MD, Neubauer S: Normal human left and right ventricular and left atrial dimensions using steady state free precession magnetic resonance imaging. J Cardiovasc Magn Reson 2005, 7:

22 7. Karamitsos TD, Hudsmith LE, Selvanayagam JB, Neubauer S, Francis JM: Operator induced variability in left ventricular measurements with cardiovascular magnetic resonance is improved after training. J Cardiovasc Magn Reson 2007, 9: Bloomer TN, Plein S, Radjenovic A, Higgins DM, Jones TR, Ridgway JP, Sivananthan MU: Cine MRI using steady state free precession in the radial long axis orientation is a fast accurate method for obtaining volumetric data of the left ventricle. J Magn Reson Imaging 2001, 14: Alfakih K, Plein S, Thiele H, Jones T, Ridgway JP, Sivananthan MU: Normal human left and right ventricular dimensions for MRI as assessed by turbo gradient echo and steady- state free precession imaging sequences. J Magn Reson Imaging 2003, 17: Natori S, Lai S, Finn JP, Gomes AS, Hundley WG, Jerosch- Herold M, Pearson G, Sinha S, Arai A, Lima JA, Bluemke DA: Cardiovascular function in multi- ethnic study of atherosclerosis: normal values by age, sex, and ethnicity. AJR Am J Roentgenol 2006, 186:S Salton CJ, Chuang ML, O'Donnell CJ, Kupka MJ, Larson MG, Kissinger KV, Edelman RR, Levy D, Manning WJ: Gender differences and normal left ventricular anatomy in an adult population free of hypertension. A cardiovascular magnetic resonance study of the Framingham Heart Study Offspring cohort. J Am Coll Cardiol 2002, 39: Thiele H, Paetsch I, Schnackenburg B, Bornstedt A, Grebe O, Wellnhofer E, Schuler G, Fleck E, Nagel E: Improved accuracy of quantitative assessment of left ventricular volume and ejection fraction by geometric models with steady- state free precession. J Cardiovasc Magn Reson 2002, 4:

23 13. Dulce MC, Mostbeck GH, Friese KK, Caputo GR, Higgins CB: Quantification of the left ventricular volumes and function with cine MR imaging: comparison of geometric models with three- dimensional data. Radiology 1993, 188: Hergan K, Schuster A, Fruhwald J, Mair M, Burger R, Topker M: Comparison of left and right ventricular volume measurement using the Simpson's method and the area length method. Eur J Radiol 2008, 65: Sievers B, Brandts B, Franken U, Trappe HJ: Single and biplane TrueFISP cardiovascular magnetic resonance for rapid evaluation of left ventricular volumes and ejection fraction. J Cardiovasc Magn Reson 2004, 6: Kwong RY: Cardiovascular Magnetic Resonance Imaging (Contemporary Cardiology). 1 edn: Humana Press; Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986, 1: Moon JC, Lorenz CH, Francis JM, Smith GC, Pennell DJ: Breath- hold FLASH and FISP cardiovascular MR imaging: left ventricular volume differences and reproducibility. Radiology 2002, 223: Papavassiliu T, Kuhl HP, Schroder M, Suselbeck T, Bondarenko O, Bohm CK, Beek A, Hofman MM, van Rossum AC: Effect of endocardial trabeculae on left ventricular measurements and measurement reproducibility at cardiovascular MR imaging. Radiology 2005, 236: Weinsaft JW, Cham MD, Janik M, Min JK, Henschke CI, Yankelevitz DF, Devereux RB: Left ventricular papillary muscles and trabeculae are significant determinants of cardiac MRI volumetric measurements: effects on clinical standards in patients with advanced systolic dysfunction. Int J Cardiol 2008, 126:

24 21. Maceira AM, Prasad SK, Khan M, Pennell DJ: Normalized left ventricular systolic and diastolic function by steady state free precession cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2006, 8: Fieno DS, Jaffe WC, Simonetti OP, Judd RM, Finn JP: TrueFISP: assessment of accuracy for measurement of left ventricular mass in an animal model. J Magn Reson Imaging 2002, 15: Francois CJ, Fieno DS, Shors SM, Finn JP: Left ventricular mass: manual and automatic segmentation of true FISP and FLASH cine MR images in dogs and pigs. Radiology 2004, 230: Grothues F, Smith GC, Moon JC, Bellenger NG, Collins P, Klein HU, Pennell DJ: Comparison of interstudy reproducibility of cardiovascular magnetic resonance with two- dimensional echocardiography in normal subjects and in patients with heart failure or left ventricular hypertrophy. Am J Cardiol 2002, 90:

25 Figure legends Figure 1. Basal short- axis slice selection. A and B. The basal LV slice, and how much of it to include, can be difficult to determine from short axis images alone. At the base of the heart, slices were considered to be within the LV at end- diastole (A) and end- systole (B) if the cavity was surrounded by 50% or more of ventricular myocardium. If the basal slice contained both ventricular and atrial myocardium, contours were drawn up to the junction and joined by a curved line through the blood pool. C and D. With the semi- automated analysis, mitral (and aortic) valve positions were identified at end- diastole (C) and systole (D) in all 3 long- axis images allowing valve planes to be tracked through the cardiac cycle (only horizontal long- axis image shown for illustration). This was integrated into a LV mesh created from the short- axis image analysis allowing automated three- dimensional identification of the LV base (and outflow tract) at end- diastole (E) and end- systole (F). Figure 2. Endocardial contouring. A. Simplified manual contouring; the compacted endocardial border is traced resulting in papillary muscles and trabeculae being included in volumetric and excluded from mass measurements. B. Detailed manual contouring; detailed tracing of the endocardium is performed manually such that papillary muscles and trabeculae are included in mass and excluded from volumetric measurements. C. Semi- automated contouring; a signal intensity thresholding tool allows detailed, semi- automated tracing of the endocardial border based on the signal intensity of myocardium. Papillary muscles and trabeculae are included in mass and excluded from volumetric measurements. Figure 3. End- systolic frame selection. Manual end- systolic frame selection is performed manually by visually identifying the frame with the smallest cavity. However often this is not uniform across short- axis slices, as in this example where basally the cavity appears smallest 25

26 in frame 14 but apically it appears smallest in frame 10. The semi- automated analysis technique automatically determines end- systole as the frame with the smallest cavity volume, by calculating cavity volume at each frame, which in this example is frame 11. Figure 4. Geometric models and formulae for the determination of LV volume (LVV). All measurements are made at end- diastole and end- systole to give EDV and ESV. A M = short axis area at the level of the base, A P = short axis area at the level of the papillary muscles, A H = long axis area in the horizontal plane, A V = long axis area in the vertical plane, L = length of the LV, D = short axis diameter of LV measured in the long axis plane Figure 5. Biplane model. Example of one of the geometric models. For the biplane model the endocardial border is traced in the horizontal long axis (top left image) and vertical long axis images at end- diastole (pictured) and end- systole. These areas are used to calculate end- diastolic and end- systolic volumes using the formula given in Figure 4. Figure 6. Bland Altman comparisons of each analysis technique for measurement of volume indices, EF and mass. Abbreviations as in Table 2. Figure 7. Interobserver and intraobserver variability of each analysis technique for measurement of volume indices, EF and mass. Interobserver and intraobserver variability (calculated using the repeatability co- efficient) is highest for measurement of mass. Intraobserver variability is less marked than interobserver variability, as would be expected. Abbreviations as in Table 2. 26

27 Figure 8. Interobserver and intraobserver variability of each geometric model for measurement of volume indices and EF. Interobserver and intraobserver variability, calculated using the repeatability co- efficient. Abbreviations as in Table 4. 27

28 Table 1. Patient characteristics, scan indications and main findings. Number (%) n = 50 Age (range 19-84) Male 34 (76%) Arrhythmia Frequent univentricular ectopics Atrial fibrillation Atrial flutter 5 (10%) Scan indication Myocardial perfusion assessment Viability assessment* Valvular assessment Aortic assessment 36 (72%) 10 (20%) 2 (4%) 2 (4%) Main scan diagnosis Ischaemic heart disease Cardiomyopathy Normal Valvular heart disease Pericardial disease Ascending aortic aneurysm Intra- cardiac mass 20 (40%) 11 (22%) 10 (20%) 4 (8%) 2 (4%) 2 (4%) 1 (2%) * includes assessment for cardiomyopathy, pericardial disease, intra- cardiac masses and myocardial viability. 28

29 Table 2. Mean + SD (range) volumetric, EF and mass measurements using each analysis technique. Auto ManDet ManSimp EDV (mls) (61-328) (69-315) (77-324) ESV (mls) (23-240) (19-245) (24-258) SV (mls) (13-150) (19-154) (21-142) EF (%) (21-81) (20-78) (19-72) Mass (g) (74-233) (52-176) (47-158) Auto semi- automated analysis, ManDet detailed manual analysis, ManSimp simplified manual analysis. EDV end- diastolic volume, ESV end- systolic volume, SV stroke volume, EF ejection fraction. 29

30 Table 3. Comparison of volumetric, EF and mass measurements using each analysis technique. Mean difference p- value 95% limits of Correlation +SD agreement coefficient (r s ) EDV (mls) Auto ManDet to Auto ManSimp < to ManDet ManSimp < to ESV (mls) Auto ManDet to Auto ManSimp < to ManDet ManSimp < to SV (mls) Auto ManDet to Auto ManSimp 8+10 < to ManDet ManSimp 7+6 < to EF (%) Auto ManDet to Auto ManSimp 7+5 < to ManDet ManSimp 6+4 < to Mass (g) Auto ManDet < to Auto ManSimp < to ManDet ManSimp 9+5 < to Mean difference (SD), significance of the difference, Bland- Altman 95% limits of agreement and correlation coefficient for the comparison of semi- automated, detailed manual and simplified manual analyses. Abbreviations as in Table 2. 30

31 Table 4. Mean + SD volumetric measurements using each geometric model. Teich ModSimp Hemi Mono Biplane Triplane EDV (mls) ESV (mls) EF (%) Teich Teicholz, ModSimp modified Simpson s, Hemi hemi- ellipse, Mono monoplane. 31

32 Table 5. Comparison of volumetric and EF measurements using each geometric model with semi- automated analysis. Mean difference p- value 95% limits of Correlation +SD agreement coefficient (r s ) EDV (mls) Auto Teich < to Auto ModSimp to Auto Hemi < to Auto Mono to Auto Biplane to Auto Triplane < to ESV (mls) Auto Teich < to Auto ModSimp < to Auto Hemi < to Auto Mono to Auto Biplane to Auto Triplane < to EF (%) Auto Teich to Auto ModSimp 9+9 < to Auto Hemi 7+7 < to Auto Mono to Auto Biplane to Auto Triplane 3+6 < to Mean difference (SD), significance of the difference, Bland- Altman 95% limits of agreement and correlation coefficient for the comparison of each geometric model with semi- automated analysis. Abbreviations as in Table 4. 32

33 33

34 34

35 35

36 36

37 37

38 38

39 39

40 40