Automatic model-based segmentation of the heart in CT images

Feasibility of automatic assessment of four-chamber cardiac function with MDCT: Initial clinical application and validation

BACKGROUND

The ability to perform a simultaneous analysis of ventricular and atrial volumes may provide clinically useful information for diagnosis and prognosis. We aimed to evaluate the feasibility and clinical value of a novel algorithm that performs fully automatic evaluation of the four cardiac chambers and myocardium from gated CT datasets.

METHODS

50 patients were studied-Group 1: 30 consecutive unselected patients, Group 2A: 10 patients after myocardial infarction and Group 2B: 10 normal controls. Fully automatic, segmentation of the heart was performed with a model-based segmentation algorithm requiring no user input other than loading the datasets. Qualitative and quantitative evaluation of segmentation quality was performed. Left ventricular (LV) and right ventricular (RV) stroke volumes (SV) were compared.

RESULTS

Overall, segmentation succeeded in all patients although 11/500 (2.2%) cardiac chambers achieved poor segmentation grading. Correlation coefficients between automatic and manually derived volumes were excellent (r>0.98) for all chambers. Bland-Altman analysis showed minimal bias (-1.0ml, 0.4ml, -1.8ml) for the LV and RV, and right atria, respectively, with mild overestimation of LV myocardial volume (5.2ml). Significant, yet consistent, overestimation of left atrial volume (23.6ml) due to inclusion of proximal pulmonary veins was observed. LV and RV ejection fraction (r=0.91 and 0.98) and SV (r=0.98 and 0.99) also correlated closely with minimal bias (<2%). Most significantly, LV SV (91.0+/-21.6ml) correlated highly with RV SV (81.7+/-18.2ml, r=0.86). Outliers could usually be explained by valvular regurgitation.

CONCLUSIONS

Fully automatic segmentation of all cardiac chambers can be achieved with high accuracy over multiple cardiac phases, enabling reliable comprehensive evaluation of four-chamber cardiac function.

Automatic intra-operative generation of geometric left atrium/pulmonary vein models from rotational X-ray angiography

Pre-procedural imaging with cardiac CT or MR has become popular for guiding complex electrophysiology procedures such as those used for atrial fibrillation ablation therapy. Electroanatomical mapping and ablation within the left atrium and pulmonary veins (LAPV) is facilitated using such data, however the pre-procedural anatomy can be quite different from that at the time of intervention. Recently, a method for intra-procedural LAPV imaging has been developed based on contrast-enhanced 3-D rotational X-ray angiography (3-D RA). These intraprocedural data now create a compelling need for rapid and automated extraction of the LAPV geometry for catheter guidance. We present a new approach to automatic intra-procedural generation of LAPV surfaces from 3-D RA volumes. Using model-based segmentation, our technique is robust to imaging noise and artifacts typical of 3-D RA imaging, strongly minimizes the user interaction time required for segmentation, and eliminates inter-subject variability. Our findings in 33 patients indicate that intra-procedural LAPV surface models accurately represent the anatomy at the time of intervention and are comparable to pre-procedural models derived from CTA or MRA.

Computational cardiac atlases: from patient to population and back

Integrative models of cardiac physiology are important for understanding disease and planning intervention. Multimodal cardiovascular imaging plays an important role in defining the computational domain, the boundary/initial conditions, and tissue function and properties. Computational models can then be personalized through information derived from in vivo and, when possible, non-invasive images. Efforts are now established to provide Web-accessible structural and functional atlases of the normal and pathological heart for clinical, research and educational purposes. Efficient and robust statistical representations of cardiac morphology and morphodynamics can thereby be obtained, enabling quantitative analysis of images based on such representations. Statistical models of shape and appearance can be built automatically from large populations of image datasets by minimizing manual intervention and data collection. These methods facilitate statistical analysis of regional heart shape and wall motion characteristics across population groups, via the application of parametric mathematical modelling tools. These parametric modelling tools and associated ontological schema also facilitate data fusion between different imaging protocols and modalities as well as other data sources. Statistical priors can also be used to support cardiac image analysis with applications to advanced quantification and subject-specific simulations of computational physiology.