Database of cardiac MR images

Images have been collected at Rouen University Hospital, as part of the clinical routine. The image database includes 754 short-axis MR images acquired on 47 patients with diagnosed pathologies, who gave written informed consent. For each patient, two volumes a total of 16 images (in average) are available at two time points: 9 images (or slices) at ED and 7 images at ES.

Cardiac MRI examinations were performed at 1.5 T (Symphony Tim®, Siemens Medical Systems, Erlangen, Germany). A dedicated eight-element phased-array cardiac coil was used. A retrospective synchronisation with a balanced steady-state free precession (bSSFP) sequence was performed for cine analysis, with repeated breath-holds of 10–15 s. All conventional planes (2-, 3- and 4-chambers views) were acquired and a total of 8–12 contiguous cine short axis (SA) slices were performed from the base to the apex of the ventricles. ED and ES were located on a mid-ventricular SA slice, and visually defined as the image with the largest and the smallest cavity respectively. Sequence parameters were as follows: TR = 50 ms; TE = 1.7 ms; flip angle = 55°; slice thickness = 7 mm; matrix size = 256×216; Field of view = 360–420 mm; 20 images per cardiac cycle; spatial resolution: 0.7mm per pixel. Cardiac images have been zoomed and cropped to a 256x216 (or 216x256) pixel ROI, leaving the LV visible for joint ventricle segmentation, if necessary.

Reference standard

Manual segmentation of the endocardium and epicardium was performed by a cardiac radiologists, with the convention that trabeculae and papillary muscles were included in the RV cavity. 

Evaluation methods

An overlap measure (Dice metric) is computed based on the pixel labeling as the result of a segmentation algorithm. Hausdorff distance between the automatic and the manual contours is also computed.

Based on the ED and ES myocardial surface segmentation, the following global RV functions are also assessed: RV ED and ES volumes (ml), ejection fraction and RV ED Mass (g).

References

J. Caudron, J. Fares, V. Lefebvre, P.-H. Vivier, C. Petitjean, and J.-N. Dacher. Cardiac MR Assessment of Right Ventricular Function in Acquired Heart Disease: Factors of Variability, Academic Radiology, in press, 2012

J. Caudron, J. Fares, V. Lefebvre, P.-H. Vivier, C. Petitjean, and J.-N. Dacher. Diagnostic accuracy and variability of three semi-quantitative methods for assessing right ventricular systolic function from cardiac MRI in patients with acquired heart disease, European Radiology, 21(10), 2111-20, 2011

S K Warfield, K H Zou, and W M Wells. 2004. Simultaneous truth and performance level estimation (STAPLE): an algorithm for the validation of image segmentation. IEEE Trans Med Imaging 23 (7):903-21.