Medical image computingMyocardial delineation via registration in a polar coordinate system1☆
Section snippets
Automated myocardial delineation
When designing an image analysis technique, it makes intuitive sense to use all of the available information to help perform the task. There are essentially three types of information that can be used in this process: anatomical information, acquisition information, and image information. Anatomical information could include, for example, the fact that the left ventricular myocardium is expected to be a closed ring of muscle encompassing the left ventricular blood pool, or that one would expect
Data
Short axis electrocardiogram retrospectively gated steady state free precession SENSE (33) images (Fig 1, top row) were obtained in 10 patients undergoing cardiac MRI for the investigation of ischemic heart disease. The images had eight to nine contiguous slices, taken in blocks of three slices over three breath holds, imaged with slice thickness 8 mm–10 mm, field of view 350 mm × 344 mm–390 mm × 390 mm, acquisition matrix 192 × 192 with 120% phase encode direction sampling, reconstructed to
Results
Figure 2 shows examples of contours produced via registration propagation of segmentations. Examples of the endo- and epicardial volumes (produced by summing the in-slice volumes) for an example patient, are shown in Figure 3, Figure 4. The traces produced via registration evolve in a smooth and physically plausible manner as opposed to the erratic evolution of the manually produced traces. This suggests that, subject to the initial delineation, segmentation propagation via registration
Discussion
We have described a new technique for analysis of left ventricular function from short axis cine cardiac MR images. The technique delineates the epicardial and endocardial boundaries by non-rigid registration of the images across the phases of the cardiac cycle, and propagation of manually delineated boundaries at one phase using the calculated deformation field. The proposed technique is affected by the following issues:
Conclusion
A novel approach to the analysis of cardiac magnetic resonance images has been introduced. Ten sets of patient images were resampled into a polar coordinate system and initial manual delineations were propagated following non-rigid registration of the resampled images. In total, contours were propagated to 1,052 images. Expert delineation time was reduced from around 3 hours for manual delineation to around 9 minutes for the described technique. The resulting propagated contours were then
Acknowledgements
We are grateful to Philips Medical Systems Nederland B.V. Medical Imaging Information Technology-Advanced Development for funding this work, to those at the Imaging Sciences Division for their assistance, and in particular to P. Batchelor for many invaluable discussions.
References (38)
- et al.
Cardiac MRIrecent progress and continued challenges
J Magn Reson Imaging
(2002) - et al.
Cardiac magnetic resonance imaginga “one-stop-shop” evaluation of myocardial dysfunction
Curr Opin Cardiol
(2002) - et al.
Assessment of left ventricular volume and mass by cine magnetic resonance imaging in patients with anterior myocardial infarction intra-observer and inter-observer variability on contour detection
Int J Card Imaging
(1996) - et al.
A model-based four-dimensional left ventricular surface detector
IEEE Trans Med Imaging
(1991) - et al.
Knowledge-based system for boundary detection of four-dimensional cardiac magnetic resonance image sequences
IEEE Trans Med Imaging
(1993) - et al.
Automated myocardial edge detection from breath-hold cine-MR imagesevaluation of left ventricular volumes and mass
Magn Reson Imaging
(1994) - et al.
Segmentation of cardiac cine MR images for extraction of right and left ventricular chambers
IEEE Trans Med Imaging
(1995) - et al.
Methods of graph searching for border detection in image sequences with applications to cardiac magnetic resonance imaging
IEEE Trans Med Imaging
(1995) - et al.
Adaptive blood pool segmentation in three-dimensionsapplication to MR cardiac evaluation
J Magn Reson Imaging
(1996) - et al.
An automatic contour extraction algorithm for short-axis cardiac magnetic resonance images
Med Phys
(1996)
Automatic assessment of cardiac function from short-axis MRIprocedure and clinical evaluation
Magn Reson Imaging
Automated detection of the left ventricular region in magnetic resonance images by fuzzy C-means model
Int J Card Imaging
A virtual exploring robot for adaptive left ventricle contour detection in cardiac MR images
Comparison between manual and semiautomated analysis of left ventricular volume parameters from short-axis MR images
J Comput Assist Tomogr
Automatic cardiac segmentation using edge detection by tissue classification in pixel neighborhoods
Magn Reson Med
Geodesic deformable models for medical image analysis
IEEE Trans Med Imaging
Determination of normal regional left ventricular function from cine-MR images using a semi-automated edge detection method
Magn Reson Imaging
Clinical evaluation of an automated boundary tracking algorithm on cardiac MRI images
Int J Card Imaging
Mathematical modeling of the heart using magnetic resonance imaging
IEEE Trans Med Imaging
Cited by (0)
- ☆
Funding provided by EasyVision Advanced Development, Medical Information Technology, Philips Medical Systems, Best, The Netherlands.