Application of a new segmentation tool based on interactive simplex meshes to cardiac images and pulmonary MRI data.

RATIONALE AND OBJECTIVES: Medical image segmentation is still very time consuming and is therefore seldom integrated into clinical routine. Various three-dimensional (3D) segmentation approaches could facilitate the work, but they are rarely used in clinical setups because of complex initialization and parametrization of such models. MATERIALS AND METHODS: We developed a new semiautomatic 3D-segmentation tool based on deformable simplex meshes. The user can define attracting points in the original image data. The new deformation algorithm guarantees that the surface model will pass through these interactively set points. The user can directly influence the evolution of the deformable model and gets direct feedback during the segmentation process. RESULTS: The segmentation tool was evaluated for cardiac image data and magnetic resonance imaging lung images. Comparison with manual segmentation showed high accuracy. Time needed for delineation of the various structures could be reduced in some cases. The model was not sensitive to noise in the input data and model initialization. CONCLUSIONS: The tool is suitable for fast interactive segmentation of any kind of 3D or 3D time-resolved medical image data. It enables the clinician to influence a complex 3D-segmentation algorithm and makes this algorithm controllable. The better the quality of the data, the less interaction is required. The tool still works when the processed images have low quality.

The medical imaging interaction toolkit.

Thoroughly designed, open-source toolkits emerge to boost progress in medical imaging. The Insight Toolkit (ITK) provides this for the algorithmic scope of medical imaging, especially for segmentation and registration. But medical imaging algorithms have to be clinically applied to be useful, which additionally requires visualization and interaction. The Visualization Toolkit (VTK) has powerful visualization capabilities, but only low-level support for interaction. In this paper, we present the Medical Imaging Interaction Toolkit (MITK). The goal of MITK is to significantly reduce the effort required to construct specifically tailored, interactive applications for medical image analysis. MITK allows an easy combination of algorithms developed by ITK with visualizations created by VTK and extends these two toolkits with those features, which are outside the scope of both. MITK adds support for complex interactions with multiple states as well as undo-capabilities, a very important prerequisite for convenient user interfaces. Furthermore, MITK facilitates the realization of multiple, different views of the same data (as a multiplanar reconstruction and a 3D rendering) and supports the visualization of 3D+t data, whereas VTK is only designed to create one kind of view of 2D or 3D data. MITK reuses virtually everything from ITK and VTK. Thus, it is not at all a competitor to ITK or VTK, but an extension, which eases the combination of both and adds the features required for interactive, convenient to use medical imaging software. MITK is an open-source project (www.mitk.org).

Visualization and attributation of vascular structures for diagnostics and therapy planning.

In various medical fields vascular structures have to be examined with usually two-dimensional views which present imaging techniques produce. The interpretation of the data can be supported by 3-dimensional visualization techniques. The further analysis requires often the attributation of the particular functional or anatomical entities. To attribute these interactively we developed two different visualization strategies. In the first one the shape of the structures is modelled with OpenGL achieving very fast response times, most notably during the navigation. The second strategy, the direct rendering of the volume, benefits from the accurate reproduction of the vascular structures. Although the rendering needs much more time, the strategy provides similar response times for the attributation. Thus, the strategies complement one another.

Virtual reality in 3D echocardiography: dynamic visualization of atrioventricular annuli surface models and volume rendered Doppler-ultrasound.

Knowledge about annuli shape and blood flow patterns, both optimally assessed by transesophageal 3D Doppler echocardiography, can be used in computer assisted surgical planning of heart valve reconstruction. Moreover, information about the individual shape of the annulus anatomy can guide the design of annular prostheses. The problem is that the annulus cannot be easily differentiated from the valve and the myocardium with standard visualization methods. We have developed a nearly automatic method for annulus segmentation. The algorithm provides the annulus shape in a symbolic description, which can be used for surface visualization. Best results to visualize the blood flow from the Doppler signal and the myocardial morphology are obtained by volume rendering. A hybrid visualization technique combining surface rendering and volume rendering enables to dynamically visualize the surface rendered annulus combined with a volume rendered 3D (plus time) reconstruction of either backscatter (morphology) and Doppler information (in original color coding), or together with backscatter only or Doppler only. Visualization of annuli structures combined with blood flow and general myocardial morphology provides a new tool to analyze heart diseases.

Improved correlation of histological data with DCE MRI parameter maps by 3D reconstruction, reslicing and parameterization of the histological images.

Due to poor correlation of slice thickness and orientation, verification of radiological methods with histology is difficult. Thus, a procedure for three-dimensional reconstruction, reslicing and parameterization of histological data was developed, enabling a proper correlation with radiological data. Two different subcutaneous tumors were examined by MR microangiography and DCE-MRI, the latter being post-processed using a pharmacokinetic two-compartment model. Subsequently, tumors were serially sectioned and vessels stained with immunofluorescence markers. A ray-tracing algorithm performed three-dimensional visualization of the histological data, allowing virtually reslicing to thicker sections analogous to MRI slice geometry. Thick slices were processed as parameter maps color coding the marker density in the depth of the slice. Histological 3D reconstructions displayed the diffuse angioarchitecture of malignant tumors. Resliced histological images enabled specification of high enhancing areas seen on MR microangiography as large single vessels or vessel assemblies. In orthogonally reconstructed histological slices, single vessels were delineated. ROI analysis showed significant correlation between histological parameter maps of vessel density and MR parameter maps (r=0.83, P=0.05). The 3D approach to histology improves correlation of histological and radiological data due to proper matching of slice geometry. This method can be used with any histological stain, thus enabling a multivariable correlation of non-invasive data and histology.