Using anatomical knowledge expressed as fuzzy constraints to segment the heart in CT images

Abstract

International audienceSegmenting the heart in medical images is a challenging and important task for many applications. In particular, segmenting the heart in CT images is very useful for cardiology and oncological applications such as radiotherapy. Although the majority of methods in the literature are designed for ventricle segmentation, there is a real interest in segmenting the heart as a whole in this modality. In this paper, we address this problem and propose an automatic and robust method, based on anatomical knowledge about the heart, in particular its position with respect to the lungs. This knowledge is represented in a fuzzy formalism and it is used both to define a region of interest and to drive the evolution of a deformable model in order to segment the heart inside this region. The proposed method has been applied on non-contrast CT images and the obtained results have been compared to manual segmentations of the heart, showing the good accuracy and high robustness of our approach

Introducing Shape Constraint via Legendre Moments in a Variational Framework for Cardiac Segmentation on Non-Contrast Ct Images

“…Similarity, sensitivity and specificity indices are computed and reported in Table 2. We also compared our results with those obtained by the method of Moreno (Moreno et al, 2008). We globally obtain better results than those obtained by Moreno et al Differences between the results of the two methods are illustrated in Figures 7(b) and 8.…”

“…In particular, the shape constraint allows us to achieve a good separation between the heart and surrounding organs (liver, aorta), improving the initial fuzzy region competition model. When compared to another method (Moreno et al, 2008) using structural knowledge (but no shape information) the results are also improved. This framework could be extended in a sequential way to segment other organs in the thorax like the aorta.…”

“…We performed several experiments for different parameters and finally we fixed τ = 1.4 and γ = 0.8 for all tests summarized in Table 2. (Moreno et al, 2008). From a pre-segmentation of the lungs, a region of interest (ROI) around the heart is built as the bounding box of lungs elongated at the bottom to insure that the heart is completely inside.…”

“…Unfortunately, this method does not apply to non-contrast CT. In (Moreno et al, 2008), the segmentation is constrained using fuzzy representations of anatomical knowledge about the position of the heart in the thorax and with respect to the lungs. This leads to a good robustness but to an average similarity index when compared to manual segmentation of 0.74, which might be too limited for radiotherapy applications.…”

3D Angiographic Image Segmentation

“…The readers who are interested in other works related to mathematical morphologybased segmentation methods of 3D angiographic data can also refer to the following articles [21,2,20,1]. More generally, besides the use of morphological and geometric knowledge [9,13], other solutions dealing with the integration of topological knowledge [6,5,7], relational knowledge [3,8], or even the use of temporal information [4] (see also the previous chapter of this book) for the segmentation of medical images, have been investigated during the last years, and have already led to quite interesting results.…”