Automatic segmentation of the left ventricle from SPECT images of rats

Congress:

ESCR 2014

Poster Number:

422

Type:

Scientific Poster

Keywords:

Cardiac, SPECT, Computer Applications-3D, Computer Applications-General, Ischaemia / Infarction

Authors:

P.-M. Jodoin¹, M. Havaei¹, A. Oudot², A. Lalande²; ¹Sherbrooke/CA, ²Dijon/FR

DOI:

DOI-Link:

https://dx.doi.org/

Fig. 1: Spherical-cylindrical model of the left ventricle.

Fig. 2: SPECT scan of a rat's LV in long axis view with the T-bar shape obtained from...

Fig. 3: Image of the figure 2 displayed in the S-C space. The red and green points are...

Fig. 4: Automatic segmentation of the endocardical border (in red) and the epicardial...

Fig. 5: 3D modelling of the segmentation of the whole data.

Purpose

Segmentation of the left ventricle (LV) of rats in SPECT images is a challenging task due to the low resolution compared with the size of the studied organ. We propose a fast automated 3D segmentation method based on graph-cut. The segmentation is performed in a particular space that model the real anatomical shape of the LV.

Methods and Materials

Making the assumption that the LV has an anatomical U shape [Ale 10, Ben 96], 3D sampling of the LV cavity is made in a spherical-cylindrical (S-C) coordinate system. Conversion from Euclidean space to the S-C space is done by dividing the LV into a spherical part (the apex) and a cylindrical part (the base) (Fig. 1). The only manual interactions are for the user to position the base and the apex of a T-bar shape as shown in Fig.2. Then, the anatomical U-shape of...

Results

The figure 4 shows an example of the automatic segmentation of the LV borders according to a long axis orientation. A 3D modelling from the whole data is displayed on fig. 5. By comparing the results obtained with our method and the Flowquant software (with a validated segmentation method [Kle 10]), there is no significant difference for the calculation of the diastolic volume (r=0.99; y=0.83 x ± 0.3), of the systolic volume (r=0.94; y=1.02 x ± 0.09) and of the ejection fraction (r=0.94; y=1.03 x...

Conclusion

Our proposed method allows a fast determination of the LV parameters on rats with a high robustness and minimal user-interaction. Compared with a validated segmentation software, although results are similar, our approach is substantially faster, requires few parameter adjustment and few user-interactions.

References

[Ale 10] M. Alessandrini, T. Dietenbeck, D. Barbosa, J. D’hooge, O. Basset, N. Speciale, D. Friboulet, and O. Bernard, “Segmentation of the full myocardium in echocardiography using constrained level-sets”, in Computing in Cardiology 2010, pp. 409–412, 2010. [Ben 96] T. Benoit, D. Vivegnis, J. Foulon, and P. Rigo, “Quantitative evaluation of myocardial single-photon emission tomographic imaging: application to the measurement of perfusion defect size and severity”, Eur J Nucl Med, vol. 23, no. 12, pp. 1603–1612, 1996. [Boy 01] Y.Boykov, O.Veksler, and R.Zabih, “Fast approximate...