MR-Diffusion Imaging Software


MR-Diffusion Imaging Software (Health Informatics)

Software implementation of the A-GLYPH LIC algorithm

The A-GLYPH LIC algorithm allows visualization of diffusion-weighted Magnetic Resonance Imaging (dw-MRI) datasets, revealing local anisotropy features as well as global fiber architecture. The method was developed by Prof. Hans-Heino Ehricke, Mark Höller, M.Sc. and Prof. Kay Otto and is protected by patent No. DE102013213010. Here we provide a first free software implementation for physicians allowing evaluation of the method with high angular resolution dw-MRI datasets.

Overview of A-GLYPH LIC

Originally, the Line Integral Convolution (LIC) approach was designed for flow visualization by engraving a vector field’s structure onto a noise texture. LIC is essentially a filtering technique that blurs an input texture locally along a given vector field, thus providing highly correlated voxel intensities along field lines. Initially, for each voxel of the input texture, a field line is integrated over a fixed number of voxels, using the voxel as a tracking seed. The field line is used as the kernel of a convolution operator, which averages voxel intensities along the line. In A-GLYPH LIC a white noise texture is not used as originally proposed by Cabral and Leedom, rather an anisotropic spot pattern, which is continuously sampled along integral lines, is implemented. Furthermore, we use a multi-kernel approach allowing more than a single anisotropy direction to be visualized for each voxel. Figure 1 shows an overview of the approach adopted,depicting the most relevant processing steps and data elements.


Fig. 1: Overview of A-GLYPH LIC processing steps.

From the acquired HARDI diffusion dataset FODs are computed by spherical deconvolution. The FOD volume is used to create a high-resolution volume of glyph samples, which are used as the input pattern to a multi-kernel LIC algorithm. The resulting LIC volume is a three-dimensional gray scale texture representing regional anisotropic behavior. Additionally, a direction volume is generated, in which the averaged anisotropy direction within the LIC convolution kernel is stored as a direction vector for each voxel. This is used for directional color encoding of slices through the LIC volume. In the visualization step, LIC slices can be fused with anatomic slice images, for example from a T1 weighted dataset. By the application of volume rendering techniques, the LIC volume can be three-dimensionally visualized as a whole or after definition of a volume of interest. The details of (i) sample volume generation, (ii) multi-kernel LIC and (iii) visualization are described in:

(1) M. Höller, K.-M. Otto, U. Klose, S. Groeschel and H.-H. EHRICKE: Fiber Visualization with LIC Maps Using Multidirectional Anisotropic Glyph Samples. Int. J. Biomed. Im., Vol. 2014, Article-ID 401819 , 2014.

(2) Höller, M., Klose, U., Groeschel, S., Otto, K.-M. und Ehricke, H-H.: Visualization of MRI Diffusion Data by a Multi-Kernel LIC Approach with Anisotropic Glyph Samples. In: L. Linsen et al., Visualization in Medicine and Life Sciences III, Springer, Berlin, Heidelberg, New York, pp 157 – 176, 2016.


Fig. 2: A-GLYPH LIC image of a tumor patient embedded in 3D visualization of tumor surface and slices from T1 weighted dataset.

A detailed description of results from a first clinical evaluation study with patients affected by different pathologies (fiber dicplacing tumor, tumor partly infiltrating fiber tracts, selective involvement of fiber tracts, demyelinating lesion) can be found in:

(3) , Clinical Application of Fiber Visualization with LIC Maps Using Multidirectional Anisotropic Glyph Samples (A-Glyph LIC), J. Clinical Neuroradiology, Online First article, 2015.

Software download

The free fiberViewMR application program allows DICOM conformant MR diffusion images to be processed and high-quality A-GLYPH LIC slice images with sagittal, axial and coronal orientations to be generated. No parameter tuning is necessary. We recommend usage of high-angular resolution diffusion datasets with 48 directions or more.


Your computer must meet the following prerequsites to run fiberViewMR:

Windows operating system: Win64, 8 GB RAM (16 GB recommended), Java Runtime Environment (to run the installer). It is recommended to use up-to-date hardware (multi-core CPU and highend graphics card with OpenCL 1.2 driver) in order to achieve acceptable performance.


MacOSX operating system: MacOSX 10.6.0 or newer, 8 GB RAM (16 GB recommended). It is recommended to use up-to-date hardware (multi-core CPU and highend graphics card with OpenCL 1.2 driver installed) in order to achieve an acceptable performance.


Download links for fiberViewMR, Release 2.0.0:

fiberViewMR 2.0.0 for Windows: click here

fiberViewMR 2.0.0 for MacOSX: click here

Feedback and questions

For feedback and / or questions please send an e-mail to: