Materialise Mimics® CT Heart tool for heart chamber segmentation: quantitative validation
Image segmentation is the process of partitioning images to identify underlying structures. In medical images, the process represents a crucial step to identifying anatomical structures and obtaining detailed geometrical information. In cardiac applications, the structures of interest are the 4 cardiac chambers and the big arteries around the heart. Segmentation of these structures has many applications in clinical practice: estimating ventricular and atrial function (e.g. stroke volume, ejection fraction), characterizing heart defects, planning of invasive and minimally invasive therapeutic procedures, and designing and validating the geometry of various medical devices (Abadi et al. 2010; Kang et al. 2012; Zhong et al. 2013).
Manual segmentation is a time-consuming, tedious and error-prone task, therefore, automation of this process is highly desirable. Mimics® Innovation Suite (Materialise N.V., Belgium) provides a so-called CT-heart tool since version 18. The CT-heart tool is a semi-automatic tool that requires the users to indicate labels for each of the structure of interest and to indicate thresholds for the gray-value range to be taken into consideration. With these inputs, the tool is able to provide a high-quality segmentation of the heart chambers and the vessels from medical images obtained using computed tomography (CT).
The new CT Heart tool was designed for Mimics® Innovation Suite 22.0. It provides a high-quality and completely automatic segmentation of the heart chambers and the vessels. In this paper, we evaluated the performance of the new CT heart tool against the performance of the previous version and against another automatic method reported in literature (Yang et al. 2018). The results show that the tool was able to provide a high-quality segmentation of all the main structures of the heart. In addition, the algorithm was much faster than the method reported in literature, while it required a processing time compared to the old semi-automatic CT-heart tool.