3D freehand ultrasound of an abdominal aneurism, fused with a CT scan of the same patient.

 

ImFusion Suite Medical Imaging Software

ImFusion Suite provides advanced visualization and processing of 2D, 3D and 4D (i.e. 3D over time) medical data sets paired with user-friendly and workflow-specific interaction. A particular focus is on seamless multi-modality fusion of ultrasound (US), CT and MRI data, as well as efficient and intelligent GPU-based processing. Plugins are available for performing advanced image analysis for a variety of clinical applications. This includes automatic multi-modal image registration (e.g. for neurosurgical of prostate US-MRI fusion), freehand ultrasound processing (general 3D compounding, vascular analysis for AAA, carotid artery plaque), interactive and automatic segmentation, classification and machine learning algorithms. Another component provides iterative cone-beam computed tomography (CBCT) reconstruction, calibration and motion compensation. 3D visual reconstruction by one or multiple Kinect-like sensors is offered by our RGB-D module.

We aim at advancing the state of the art by developing superior medical image computing technology. At the same time, this technology shall be well packaged and usable by both engineers and researchers, as well as technology-oriented clinicans.

 

Data Supported
  • Modalities: CT, CBCT, MRI, ultrasound (including 4D formats), others/secondary storage
  • DICOM (single- and multi-frame files, graphical browser, DICOMDIR, PACS connectivity)
  • MetaImage Format (mhd, mha)
  • Tool dialog to load raw binary files with arbitrary type and dimension
  • 2D image data, including 16-bit PNG/TIFF/JPEG, videos
  • 3D Surface meshes (STL, PLY, OBJ)
  • Own XML-based workspaces, allowing for reproducible processing and display

Visualization
  • Standard and oblique multi-planar-reformating (MPR), curved MPR
  • Fast GPU shaded volume rendering, including DRR, MIP, iso-surface modes
  • Flexible view layouts including side-by-side for fusion, larger focus view
  • Blending and fusion of arbitrary 2D and 3D/4D image data from different modalities
  • Annotations natively residing in 3D coordinates (labels, lines, angles, curves, meshes etc.)

Registration
  • Interactive registration with landmarks, visual superimposition, side-by-side, transformation editor
  • Automatic image-based registration with rigid, affine and free-form deformation models
  • Our signature multi-modal registration (CT, MRI, US) with 3D patch-based LC2 similarity
  • Fast and global distance volume based surface registration

Segmentation
  • Manual labeling of 2D and 3D data sets, thresholding methods, fast marching
  • Graph-based semi-automatic segmentation of any anatomy in 2D or 3D, iterative refinement
  • Conversion of segmentation result to masks, contours, meshes with chosen smoothing
  • Dedicated segmentation of tubular structures and wires
  • Smart-spline interactive curve fitting

3D Freehand Ultrasound
  • High-quality live compounding of tracked freehand ultrasound data
  • Image-based algorithm for precise sensor-to-probe calibration and registration
  • Extended field-of-view ultrasound reconstruction from matrix-array volumes
  • Simulation of ultrasound from CT or MRI, including refraction, speckle and other physical properties

Machine Learning
  • Deep Learning library integration for state-of-the-art machine learning capabilities
  • Powerful Random Forest implementation with features computed on-the-fly for large problems
  • Interactive image features editor to assemble learning-based algorithms on arbitrary data
  • Support for both classification and regression applications
  • Allows for fully automated image analysis, especially when combined with registration / segmentation

Cone-Beam CT
  • Fast iterative reconstruction on the GPU, including ROI/truncation handling, statistical priors
  • Tools for both geometric and radiometric cone-beam device calibration
  • Automatic detection of unexpected patient or device motion, compensation in reconstruction

3D Surface Scanning
  • Real-time GPU-based reconstruction of surfaces, e.g. patient torso/extremities from depth sensors
  • Mesh post-processing for 3D printing (closing holes, reducing mesh complexity, ...)
  • Multi-sensor support for instant whole-object acquisition

Vascular Ultrasound

ImFusion Suite supports automatic segmentation and quantification of vascular ultrasound for instance for carotid plaque or AAA applications. Learning-based classification detects vascular structures in 3D with the push of a button. Alternatively, in the case of multiple or hardly visible structures, a single-click initialization may be performed. After the segmentation, statistics on the vessel geometry are computed and plotted, the resulting 3D surface can be exported to standard mesh formats.

This work is featured in the piur imaging system.

Demo Video for Carotid Artery Analysis

Prostate Fusion

Our powerful registration algorithms allow to fuse MRI and freehand ultrasound of the prostate for better cancer screening or targeted biopsies. To deal with this difficult and highly deformed imaging data, a number of image analysis technologies are paired with sophisticated fused visualization and interaction.

A variety of tools allow for calibration, registration and 3D reconstruction of erroneous freehand ultrasound acquisitions. Definition of anatomical landmarks together with tools for blending, superimposition and side-by-side visualization enable fusion of e.g. trans-rectal ultrasound acquisitions with an MRI scan. Henceforth, image-based registration using rigid, affine and free-form deformation models uses all available anatomical clues for a robust final alignment of the data despite its difficult appearance and geometrical inconsistency.

Neuro Navigation

ImFusion Suite supports registering 3D ultrasound data to MRI volumes for instance for brain-shift correction during neuro-surgery. In general, both registration of different ultrasound acquisitions throughout the procedure, as well multi-modal registration to different phases of an MRI scan enable optimal use of pre-operative information during neuro-surgery, as well as putting follow-up validation into context.

Demo Video for neuro US-MRI registration

This video from our MICCAI 2013 conference presentation in Nagoya demonstrates fully automatic (global) registration of intra-operative ultrasound to MRI, while showing the best configuration that has been found during the processing. In a second step, a deformable refinement aligns the lesion with the more recent ultrasound acquisition to match the intra-operative setting.
The full video of our talk is available online, as well as the corresponding publication. A further journal publication arose from our cooperation with the CAMP group of Prof. Navab in the context of US-MRI registration.

Cone-Beam CT

The CT module in ImFusion Suite contains algorithms for rapid iterative reconstruction on the GPU, in both OpenGL and OpenCL implementations. This includes ROI/truncation handling for scanning large objects, and statistical priors for low-dose protocols. Besides, we have tools for both geometric and radiometric cone-beam device calibration, which may help device manufactures to setup own imaging systems or experiments. Last but not least, novel algorithms for automatic detection of unexpected patient or device motion is a unique feature in our software, allowing to compensate for motion during the reconstruction. This technology was presented in a publication at the 2013 Fully 3D conference.

3D Freehand Ultrasound

ImFusion Suite offers extensive support for working with both 2D and 3D freehand ultrasound. Tracking information is used when available, otherwise image-based reconstruction is possible.

A variety of proprietary formats are supported to load ultrasound videos/clips alongside tracking information. Modules for image-based probe-to-sensor calibration and slice-to-volume registration allow to properly figure out the geometrical attributes. We have developed a signature on-the-fly fast compounding on the GPU (based on our previously published work), which allows to visualize cross-section in any orientation, even while geometric parameters are still being optimized. A variety of different forward and backward compounding techniques then reconstruct high-quality volumes out of the freehand slices. At this point, our general modules for fusion and multi-modal registration kick in. A demonstration of the combination of learning-based initialization, mono-modal ultrasound extended-field-of-view reconstruction, and multi-modal registration, can be found in our 2014 publication at the VCBM workshop.

Within the ImFusion framework, we have also developed a fast GPU-based ultrasound simulation algorithm which is able to synthesize medical US images including the tissue-specific speckle pattern and anisotropic ultrasound artifacts. By following the path of rays through a 3D volume and considering the acoustic properties of different tissues, this simulator can create US imaging artifacts such as refractions, reverberation, range distortion, and mirroring in the order of a few seconds. Since the algorithm is fast and yet capable of simulating many ultrasound imaging artifacts, it can boost other medical computing applications such as multi-modal image registration, medical ultrasound training, and learning tissue acoustic properties.

3D Surface Acquisition

ImFusion Suite allows you to scan the surface of human anatomy (extremities, torso etc.) for taking measurements and further analysis such as assessing surgical outcome. A number of RGB-D sensors such as the Microsoft Kinect and Asus Xtion are supported. The surface reconstruction is performed in real-time with a fast and robust reconstruction algorithm. A variety of mesh post-processing features allow for high-quality 3D printing (closing holes, reducing mesh complexity, smoothing, picking the largest component, cropping etc.). This technology is featured in our consumer product RecFusion and has undergone significant improvements and refinements since its initial release. We now also feature multi-sensor support for instant whole-object acquisitions, well suited for diverse medical scenarios.

Below are a number of selected clinical application areas where the ImFusion Suite software framework is successfully used.

 

Vascular Ultrasound

ImFusion Suite supports automatic segmentation and quantification of vascular ultrasound, e.g. for carotid plaque or AAA applications.

       

Prostate Fusion

Our powerful registration algorithms allow to fuse MRI and freehand ultrasound of the prostate for better cancer screening or targeted biopsies.

     

Neuro Navigation

ImFusion Suite supports registering 3D ultrasound data to MRI volumes for instance for brain-shift correction during neuro-surgery.

 

Cone-Beam CT

ImFusion Suite offers fast iterative cone-beam reconstruction algorithms as well as methods for geometric and radiometric calibration. Reconstructed volumes are exported as standard DICOM files.

     

3D Freehand Ultrasound

ImFusion Suite contains extensive support for working with both 2D and 3D freehand ultrasound. Tracking information is used when available, otherwise image-based reconstruction is possible.

 

3D Surface Acquisition

ImFusion Suite allows you to scan the surface of human anatomy (extremities, torso) for taking measurements and further analysis such as assessing surgical outcome. Multi-sensor setups are also supported.

Our software is currently not certified as a medical product, i.e. it has neither FDA approval nor does it bear a CE marking. ImFusion Suite is available for 64-bit Windows (7, 8.1) and Linux platforms. Upon request we are also able to provide OS-X and 32-bit installers.

  • ImFusion Suite (Display & Processing, Registration, Segmentation)
  • Machine Learning Module
  • 3D Ultrasound Module
  • Cone-beam CT Module
  • RGB-D Reconstruction Module

ImFusion Suite and each of the extension modules is also available as a software development kit (SDK) version. This enables researchers and engineers to write own plugins for ImFusion Suite and directly interface our image analysis modules using the C++ language (Microsoft Visual Studio 2012 and 2013, GCC and Clang compilers).

 

Please This email address is being protected from spambots. You need JavaScript enabled to view it. for pricing and detailed conditions!