Visualization & Exploration of Perfusion Data

Perfusion data are dynamic medical image data which characterize the regional blood flow in tissue. We focus on the visualization and exploration of perfusion data for diagnostic purposes in three major application areas: ischemic stroke diagnosis, breast tumor diagnosis and the diagnosis of coronary heart disease.

For a reliable diagnosis, the multi-dimensional space of perfusion parameters needs to be explored. Therefore, we aim at different exploration and visualization techniques for an improved diagnosis. Furthermore, visual analytics approaches will be developed and integrated in the diagnosis frameworks.

This project is a subproject of the DFG funded Scalable Visual Analytics project.

A short summary of our current field of research:

Visual Analytics for Perfusion Image Data
Multi-parameter Visualization for the Exploration of Perfusion Image Data

For Interested Students

If you are interested in working in this versatile and challenging topic feel free to visit our student offers subpage or contact us directly (see below). We offer “Hiwi”-jobs as well as “Laborpraktika / IT-Projekte” and various topics for your Bachelor or Master's Thesis/Diploma.

Contact

Name	Dipl.-Ing. Sylvia Glaßer Dr. Steffen Oeltze
Adress	Otto-von-Guericke-Universität Magdeburg Fakultät für Informatik / ISG, PSF 4120 39016 Magdeburg
Phone	(+49-391) 67-1 25 27
Fax	(+49 391) 67-1 11 64

Images

Loading in progress.

Tissue affected by an ischemic stroke is characterized by a delayed (high Time-To-Peak) and diminished (low Integral) blood supply.

The contrast agents wash-out velocity is color-coded and projected through the lens in the context of the original perfusion data.

Synchronized lenses support the comparison of both hemispheres. The initially defined lens is mirrored on a relocatable line of symmetry.

The rightmost image represents an integrated visualization of the four parameter maps by means of color icons (template shown as inset).

One glyph represents 4x4 voxels in order to gain screen space for glyph drawing. The user may interactively adjust this ratio.

The triangular glyphs resemble the original concentration-time curves generated by plotting contrast agent concentration over time.

A parameter sub-domain has been defined in a feature specification component. The glyph display is restricted to the corresponding tissue.

Focus and near-focus regions (bluish rectangles) have been defined (inset) resulting in different degrees-of-interest (DOI).

Blue, horizontally oriented glyphs indicate a diminished and decelerated perfusion caused by an infarction.

Each ring of the plot has been bisected. The resulting inner and outer rings facilitate a comparison of stress and rest perfusion (parameters).

Four slices have been acquired for evaluating the perfusion. Small bluish glyphs indicate a disturbance in the scar region (brown surface).

Division according to the AHA 17-segment model. Glyph height equals the slice thickness of the original perfusion data.

A glyph labeling, a slice of the original perfusion data and an orientation cube improve the spatial orientation.

A 3D TIC resembles the averaged signal intensity time course. The gray reference glyphs represent the entire myocardium.

The view along the left ventricular long-axis resembles a Bull's Eye Plot view. It gives an overview of the perfusion in all segments.

The right ventricle is provided as context. The left ventricular wall is emphasized at intersections with the perfusion imaging planes.

3D TICs encode the perfusion and isolines encode the transmurality of the scar. The wall thickening is color-coded on the endocardium.

Curve views show the time-intensity curves of each voxel. Brushes facilitate a differentiation of infarction core (l) and penumbra (r).

The user-defined brush resembles a curve pattern which is typical for suspicious tissue (left). Two suspicious structures are revealed (right).

Feature specification in a scatterplot (b) reveals the infarction core (a). Adding a near-focus region (d) indicates penumbral tissue (c).

Smooth gradient brushing (f) adds context information (e). Smooth brushing of PCA results (h) yields similar results (g) as compared to (c).

PCA results for 2 datasets, Heart 1/2 (a). Brushing the 1st principal component (pc) (b) reveals ischemic tissue in Heart 1 (green area (c)).

Brushing TTP and Up-Slope (f) reveals ischemic tissue in Heart 2 (green area in (e)). Brushing the 2nd pc (h) indicates preprocessing failures (g).

Brushing high intensity differences over time (b) highlights strong contrast agent wash-in (a). Analysis is focused on local region (c,d).

Identification of areas with rapid contrast agent washout (e,f). Smooth brushing (h) reveals subtle spikes along the border of Slarge (g).

Parallel coordinates plot represents perfusion parameters. Brushing reveals affected tissue. Successful therapy is indicated (3 vs. 1).

Exploiting the PCA results also reveals the affected tissue. Result is very similar as compared to perfusion parameter based analysis (1/2).

Intensity difference based (b), local (c) analysis reveals a suspicious structure (a). Its characteristics (d,e) indicate a high-grade glioma.

Curve view showing time-intensity curves of the entire dataset (d). Brushing may separate cortex (a), medulla (b), and pelvis (c).

Videos

(You can watch an enlarged version of the video by clicking on the small preview a second time.)

2008

Glyphs for encoding myocardial perfusion in 3D ( 2008 ):

Video illustrating the development and application of glyphs to encoding myocardial perfusion in 3D. The glyphs are embedded in a visualization of the heart's anatomy. (Data is courtesy of F. Grothues, Department of Cardiology, University of Magdeburg, Germany)

2007

Interactive visual analysis in breast tumor diagnosis ( 2007 ):

Video showing the interactive visual analysis of a dataset from breast tumor diagnosis. Suspicuous lesions shall be detected and characterized. (Data is courtesy of Andreas Fessel, University Hospital, Magdeburg, Germany)

2006

Integrated visualization of cardiac perfusion and anatomy ( 2006 ):

Video illustrating the integration of myocardial perfusion and ventricular as well as vascular anatomy of a patient suffering from atherosclerosis. (Data is courtesy of M. Fenchel, S. Miller and A. Seeger, Max Planck MR-center, University of Tübingen, Germany, S. Achenbach, Department of Radiology, University of Erlangen-Nürnberg, Germany, and SIEMENS Healthcare)

Publications

2011

Sylvia Glaßer, Kathrin Scheil, Uta Preim and Bernhard Preim

The File-Card-Browser View for Breast DCE-MRI Data

Bildverabeitung für die Medizin (BVM) (314-318), Lübeck, 2011