Fraunhofer Institute for Medical Image Computing

Bremen, Germany

Fraunhofer Institute for Medical Image Computing

Bremen, Germany
SEARCH FILTERS
Time filter
Source Type

Hoinkiss D.C.,Fraunhofer Institute for Medical Image Computing | Porter D.A.,Fraunhofer Institute for Medical Image Computing
NeuroImage | Year: 2017

Subject head motion is a major challenge in diffusion-weighted imaging, which requires a precise alignment of images from different time points to allow a reliable quantification of diffusion parameters within each voxel. The technique requires long measurement times, making it highly sensitive to long-term subject motion, even when head restraint is used. Current methods of data analysis rely on retrospective motion correction, but there are potential benefits to using prospective motion correction, in which motion is tracked and compensated for during data acquisition. This technique is regularly used to enhance image quality in blood-oxygen-level dependent (BOLD) imaging, but its application to diffusion-weighted imaging has been limited by the contrast variation between images acquired with different diffusion-gradient directions. This paper describes a novel approach to this topic that exploits the rotational invariance of the trace of the diffusion tensor to reduce the effect of this contrast variation, making it possible to perform a fast image registration using a least-squares cost function. This results in an image-based motion detection algorithm that can be applied in real time during data acquisition to adapt the slice position and orientation in response to subject motion. The motion detection capabilities of the technique were evaluated in a study of ten subjects with b-values up to 3000 s/mm². The resulting motion-parameter estimates were in close agreement with reference values provided by interleaved low-b-value images with a correlation coefficient of R=0.9634 for the voxel displacements measured across all subjects and b-values. The technique was also used to perform prospective motion correction on a standard clinical MRI system with b-values up to 2000 s/mm². The correction was evaluated in 3 subjects using interleaved low-b-value images, retrospective image registration using the AFNI processing package and mean diffusivity histogram analysis. Compared to acquisitions without motion correction, prospective motion correction based on pseudo-trace-weighted images was found to provide a robust method for substantially reducing the level of misregistration between volumes. In most cases, misregistrations were reduced to less than 0.2 mm of translation and 0.2° of rotation for an isotropic voxel size of 2 mm, yielding high-quality diffusion parameter maps even in the absence of head restraint and post-acquisition image registration. © 2016 Elsevier Inc.


Patz T.,Jacobs University Bremen | Preusser T.,Fraunhofer Institute for Medical Image Computing
SIAM Journal on Scientific Computing | Year: 2012

We present a model and the related discretization for phase change problems. In particular, we are focused on the evaporation of water. The governing equations inside the domains and conditions on the moving interface are derived. Afterward the numerical methods for the discretization are presented. We use a level set method to capture the interface motion and use composite finite elements (CFEs) to solve the required equations in the whole domain. CFEs are a special kind of finite elements, allowing a fast calculation, because they use structured grids and respect the geometry by adapting the basis functions in the neighborhood of an interface or at the domain boundary. For the special construction of CFEs used in this paper, we present a method to take into account Dirichlet boundary conditions on the complicated domain boundary. Also, the method used for the calculation of the interface conditions within the CFE-grid is presented. We tested the Dirichlet boundary condition method for the CFEs by solving an elliptic test problem with a precomputed right-hand side. The phase change discretization is tested by verifying the d2 law, a widely used test in the phase change and moving boundary context. With the CFE discretization, we are able to solve the full three-dimensional (3D) problem. Afterward we give a short introduction to radio-frequency ablation and present a possibility for integrating the presented phase change model into the simulation of the ablation. © 2012 Society for Industrial and Applied Mathematics.


News Article | October 28, 2016
Site: www.sciencedaily.com

Finding the ideal position for interventional needles -- as used in biopsies, for instance -- is a difficult and time-consuming process. This can now be performed automatically, using a robotic arm to place a needle guide for the doctor at the optimal insertion point. With robotic assistance, doctors need five minutes to position the needle, as opposed to 30 minutes with conventional techniques. The solution will be shown at the MEDICA trade fair in Düsseldorf from November 14 to 17, 2016 (Hall 10, Booth G05). An ultrasound shows a shadow on the liver -- but is it a tumor? Often, the only way to conclusively answer this question is to perform a biopsy, a procedure in which a doctor uses a long needle to remove a piece of the suspected tissue to be sent to a laboratory for testing. However, placing the biopsy needle with precision is far from easy. On one hand, the doctor needs to be sure of reaching the suspected tissue -- and not healthy tissue just millimeters to the side. On the other hand, the needle must not damage veins, nerve pathways, and organs such as the lungs, and cannot penetrate bony structures such as ribs. To obtain an overview, doctors begin by performing a computed tomography scan, which they use to maneuver the needle to the correct position. The same challenges arise in treatments that use needles to direct heating, cooling, or high-energy beams into the cancerous tissue, thereby destroying the tumor. Soon, precisely positioning needles will become faster thanks to a robotic arm that researchers from the Fraunhofer Institute for Manufacturing Engineering and Automa-tion IPA's Project group for Automation in Medicine and Biotechnology PAMB and the Fraunhofer Institute for Medical Image Computing MEVIS have modified specifically for this purpose. "Whereas humans struggle to position this sort of needle, it's hard to beat a robot designed for the purpose," says Andreas Rothfuss, a researcher at the PAMB. "Our system removes burdens for doctors while leaving them in control." In other words, the robot does what it does best -- locating the right path and positioning the needle guide so that there is no risk of hitting or injuring either doctor or patient. Thereafter, the doctor again takes command and inserts the needle into the tissue. "A human needs 30 minutes to position the needle, but with robot assistance this is cut down to five minutes at most," says Rothfuss. To begin the procedure, the doctor begins by performing a computed tomography scan of the patient. This time, however, the robot arm accompanies the scan using a calibration tool to determine the ideal position to target a specific point in the image. Software from Fraunhofer MEVIS analyzes the image and supports the doctor in placing the virtual needle by displaying the needle in the image. If, instead of a biopsy, the doctor is administering treatment -- seeking to destroy a tumor by applying heat, for instance -- the software simulates how the heat will spread through the tissue. The last step is to determine the number of needles and their positions required to kill off the entire tumor. Thereafter, the robot arm's calibration tool is replaced with a needle guide. The robot transports the guide to the calculated position and places it on the skin at the correct angle. However, it does not insert the needle itself: this is left to the doctor, who pushes the needle into the tissue step by step through the needle guide held in place by the robot. Less radiation exposure for doctor and patient To ensure that the needle is in the planned position, doctors take X-rays as part of the standard procedure as they insert the needle into the tissue. Here, too, the robot offers several advantages. In conventional needle insertion, doctors hold the needle in place manually, obscuring a part of the X-ray. This also exposes doctors' hands to radiation each time a monitoring image is taken. Now, the robot, impervious to radiation, can hold the needle in place with its needle guide. There is also a significant reduction in the patient's radiation exposure -- the doctor inserts the needle through the guide, eliminating needle slippage. As a result, the number of monitoring X-rays is greatly reduced. The researchers will showcase their development at the MEDICA trade fair in Düsseldorf from November 14 to 17 (Hall 10, Booth G05). The robot arm will be positioning its needle guide over a transparent plastic box complete with artificial ribs and a tumor embedded in a transparent polymer. This will allow visitors to see exactly where the needle is. Researchers hope that the system could reach the market in around three years.


News Article | October 27, 2016
Site: phys.org

An ultrasound shows a shadow on the liver – but is it a tumor? Often, the only way to conclusively answer this question is to perform a biopsy, a procedure in which a doctor uses a long needle to remove a piece of the suspected tissue to be sent to a laboratory for testing. However, placing the biopsy needle with precision is far from easy. On one hand, the doctor needs to be sure of reaching the suspected tissue – and not healthy tissue just millimeters to the side. On the other hand, the needle must not damage veins, nerve pathways, and organs such as the lungs, and cannot penetrate bony structures such as ribs. To obtain an overview, doctors begin by performing a com- puted tomography scan, which they use to maneuver the needle to the correct posi- tion. The same challenges arise in treatments that use needles to direct heating, cooling, or high-energy beams into the cancerous tissue, thereby destroying the tumor. Soon, precisely positioning needles will become faster thanks to a robotic arm that researchers from the Fraunhofer Institute for Manufacturing Engineering and Automa-tion IPA's Project group for Automation in Medicine and Biotechnology PAMB and the Fraunhofer Institute for Medical Image Computing MEVIS have modified specifically for this purpose. "Whereas humans struggle to position this sort of needle, it's hard to beat a robot designed for the purpose," says Andreas Rothfuss, a researcher at the PAMB. "Our system removes burdens for doctors while leaving them in control." In other words, the robot does what it does best – locating the right path and positioning the needle guide so that there is no risk of hitting or injuring either doctor or patient. Thereafter, the doctor again takes command and inserts the needle into the tissue. "A human needs 30 minutes to position the needle, but with robot assistance this is cut down to five minutes at most," says Rothfuss. To begin the procedure, the doctor begins by performing a computed tomography scan of the patient. This time, however, the robot arm accompanies the scan using a calibration tool to determine the ideal position to target a specific point in the image. Software from Fraunhofer MEVIS analyzes the image and supports the doctor in placing the virtual needle by displaying the needle in the image. If, instead of a biopsy, the doctor is administering treatment – seeking to destroy a tumor by applying heat, for instance – the software simulates how the heat will spread through the tissue. The last step is to determine the number of needles and their positions required to kill off the entire tumor. Thereafter, the robot arm's calibration tool is replaced with a needle guide. The robot transports the guide to the calculated position and places it on the skin at the correct angle. However, it does not insert the needle itself: this is left to the doctor, who pushes the needle into the tissue step by step through the needle guide held in place by the robot. Less radiation exposure for doctor and patient To ensure that the needle is in the planned position, doctors take X-rays as part of the standard procedure as they insert the needle into the tissue. Here, too, the robot offers several advantages. In conventional needle insertion, doctors hold the needle in place manually, obscuring a part of the X-ray. This also exposes doctors' hands to radiation each time a monitoring image is taken. Now, the robot, impervious to radiation, can hold the needle in place with its needle guide. There is also a signifi-cant reduction in the patient's radiation exposure – the doctor inserts the needle through the guide, eliminating needle slippage. As a result, the number of monitoring X-rays is greatly reduced. The researchers will showcase their development at the MEDICA trade fair in Düsseldorf from November 14 to 17 (Hall 10, Booth G05). The robot arm will be positioning its needle guide over a transparent plastic box complete with artificial ribs and a tumor embedded in a transparent polymer. This will allow visitors to see exactly where the needle is. Researchers hope that the system could reach the market in around three years.


News Article | March 1, 2017
Site: phys.org

Doctors wish to use focused ultrasound to treat tumors in moving organs, such as the liver, shown here. Credit: Fraunhofer MEVIS Focused ultrasound can effectively destroy tumor cells. Until now, this method has only been used for organs such as the prostate and uterus. At the European Congress of Radiology, Fraunhofer researchers will present a method, developed as part of the TRANS-FUSIMO EU project, that enables focused ultrasound treatment of the liver, an organ that moves while breathing. In the future, this could enable treatment of certain liver tumors in a more gentle way. Ultrasound has long served as a diagnostic method. Its application as a form of therapy treatment, however, is relatively new. In this process, ultrasound waves are highly concentrated to destroy diseased tissue, tumor cells in particular, and render them harmless. Focused ultrasound benefits patients in several ways. The therapy is completely non-invasive and can be performed without anesthesia, and there are no operation wounds. Until now, however, the method has only been approved for a limited number of indications, such as treatment of prostate cancer, bone metastases, and uterine myoma. To treat organs that move when patients breathe, the method can only be partially applied. Doctors have to rely on patients to hold their breath or put them under anesthesia, so they can control the patient's breath. The scientists working in the TRANS-FUSIMO EU project (see infobox), coordinated by the Fraunhofer Institute for Medical Image Computing MEVIS in Bremen, are following another path. They refocus the ultrasound beam to the movement of the liver to reach the tumor effectively while sparing the surrounding healthy tissue. The fundamental technology needed for the method is now ready. The researchers will present important preliminary results as part of an industry symposium at the European Radiology Congress (ECR) in Vienna on March 1. In this therapy concept, the patient lies in an MRI scanner during the procedure. Every tenth of a second, the scanner produces an image showing the current position of the liver. The ultrasound transducer, a device equipped with more than 1000 small ultrasound transmitters, sits on the patient's stomach. These can be directed so that their waves converge precisely at a point as small as a grain of rice. There, they unleash their destructive effect - the tumor cells become completely cooked. The MRI scanner controls the process, measuring the temperature in the liver and ensuring that the correct spots are sufficiently heated. Real-time software that can glance into the immediate future Project manager Sabrina Haase, mathematician at Fraunhofer MEVIS, explains the problem. "Generating an image of the liver's position every tenth of a second is not fast enough to reliably direct the ultrasound beam. This is why we developed software that can see into the immediate future and calculate the next position of the treated region." The program determines the path for the focused ultrasound waves to reach the liver tumor even when the patient moves while breathing. Developing the software was quite challenging: it must run both highly realiably and in real time. Another difficulty facing the scientists was the fact that the ribs lie in front of the liver. To prevent the beams from damaging the ribs, elements in the ultrasound transducer that would have hit the ribs were deactivated, much like blocking the holes in a showerhead which spray water in an unwanted direction. "We have completed the technical development phase and have already run preliminary tests," says Haase. In the test, a robotic arm moved a gel model back and forth in the MR scanner to simulate the liver movement inside the body. At the same time, the gel phantom was exposed to focused ultrasound, and the MRI scanner monitored the temperature distribution. "The results match our expectations," says Haase. "Now, we can pursue the next steps." The first TRANS-FUSIMO tests on patients are planned for mid-2018. Thereafter, in cooperation with an industry partner, medical product certification can be tackled. If the method proves itself, in the future, it would be possible to treat other organs that move with breathing, such as the kidney, pancreas, or even lungs.


News Article | November 14, 2016
Site: www.sciencedaily.com

Physicians have long used visual judgment of medical images to determine the course of cancer treatment. A new program package from Fraunhofer researchers reveals changes in images and facilitates this task using deep learning. The experts will demonstrate this software in Chicago from November 27 to December 2 at RSNA, the world's largest radiology meeting. Has a tumor shrunk during the course of treatment over several months, or have new tumors developed? To answer questions like these, physicians often perform CT and MRI scans. Tumors are usually evaluated only visually, and new tumors are often overlooked. "Our program package increases confidence during tumor measurement and follow-up," explains Mark Schenk from the Fraunhofer Institute for Medical Image Computing MEVIS in Bremen, Germany. "The software can, for example, determine how the volume of a tumor changes over time and supports the detection of new tumors." The package consists of modular processing components and can help medical technology manufacturers automate progress monitoring. The computer learns on its own The package is unique in its use of deep learning, a new type of machine learning that reaches far beyond existing approaches. This method is helpful for image segmentation, during which experts designate exact organ outlines. Existing computer segmentation programs seek clearly defined image features such as certain gray values. "However, this can often lead to errors," according to Fraunhofer researcher Markus Harz. "The software assigns areas to the liver that do not belong to the organ." These errors must be corrected by physicians, a process which can often be quite time-consuming. The new deep learning approaches promise improved results and should save physicians valuable time. To demonstrate their self-learning methods, Fraunhofer scientists trained the software with CT liver images from 149 patients. Results showed that the more data the program analyzed, the better it could automatically identify liver contours. A further application of the approach is image registration, in which software aligns images from different patient visits so that physicians can easily compare them. Machine learning can aid the particularly difficult task of locating bone metastases in the torso in which hip bones, ribs, and spine are visible. Currently, these metastases are often overlooked due to time constraints in clinical practice. Deep learning methods can help reliably discover metastases and thus improve treatment outcomes. Researchers focus on a combination of classical approaches and machine learning: "We wish to harness existing expertise to implement deep learning as effectively and reliably as possible," stresses Harz. Fraunhofer MEVIS builds upon years of experience in practical application: for example, the algorithms for highly precise lung image registration have been integrated into several commercial medical software applications.


News Article | November 2, 2016
Site: phys.org

Has a tumor shrunk during the course of treatment over several months, or have new tumors developed? To answer questions like these, physicians often perform CT and MRI scans. Tumors are usually evaluated only visually, and new tumors are often overlooked. "Our program package increases confidence during tumor measurement and follow-up," explains Mark Schenk from the Fraunhofer Institute for Medical Image Computing MEVIS in Bremen, Germany. "The software can, for example, determine how the volume of a tumor changes over time and supports the detection of new tumors." The package consists of modular processing components and can help medical technology manufacturers automate progress monitoring. The computer learns on its own The package is unique in its use of deep learning, a new type of machine learning that reaches far beyond existing approaches. This method is helpful for image segmentation, during which experts designate exact organ outlines. Existing computer segmentation programs seek clearly defined image features such as certain gray values. "How- ever, this can often lead to errors," according to Fraunhofer researcher Markus Harz. "The software assigns areas to the liver that do not belong to the organ." These errors must be corrected by physicians, a process which can often be quite time-consuming. The new deep learning approaches promise improved results and should save physicians valuable time. To demonstrate their self-learning methods, Fraunhofer scientists trained the software with CT liver images from 149 patients. Results showed that the more data the program analyzed, the better it could automatically identify liver contours. A further application of the approach is image registration, in which software aligns images from different patient visits so that physicians can easily compare them. Machine learning can aid the particularly difficult task of locating bone metastases in the torso in which hip bones, ribs, and spine are visible. Currently, these metastases are often overlooked due to time constraints in clinical practice. Deep learning methods can help reliably discover metastases and thus improve treatment outcomes. Researchers focus on a combination of classical approaches and machine learning: "We wish to harness existing expertise to implement deep learning as effectively and reliably as possible," stresses Harz. Fraunhofer MEVIS builds upon years of experience in practical application: for example, the algorithms for highly precise lung image registration have been integrated into several commercial medical software applications.


Jenne J.W.,Fraunhofer Institute for Medical Image Computing
Frontiers of Neurology and Neuroscience | Year: 2015

The idea to ablate brain tissue with high-intensity focused ultrasound (HIFU) in a highly precise and localized manner is relatively old. For HIFU tissue ablation, ultrasound (US) waves are concentrated to a focal point. Due to US absorption, the focal area will be heated and consequently thermally destroyed. The spatial accuracy of the non-invasive procedure and the sharp delineation of the induced tissue lesions have led to the term 'focused ultrasound surgery' (FUS). The major obstacle for HIFU ablation in the brain is the skull bone, which absorbs most of the US energy and disturbs the focused US field. The development of large-sized phased array US transducers and adaptive focusing techniques based on computed tomography images have allowed these difficulties to be overcome. With the combination of FUS and MR-imaging and MR-thermometry (MR-guided Focused Ultrasound Surgery, MRgFUS), real-time therapy guidance and control has been established. The safety, feasibility and effectiveness of transcranial MRgFUS were investigated in four initial clinical studies including 4 to 15 patients each. In the first study, which dealt with the treatment of inoperable recurrent glioblastoma, MR was used to monitor localized tissue heating, but no tissue ablation was possible due to technical restrictions of the treatment setup. With improved equipment, the precise induction of thermal lesions in the target area was achieved in studies on neuropathic pain and essential tremor. An instantaneous and persistent significant improvement of disease symptoms was observed in most patients. However, there were serious adverse effects in two cases, where intracranial hemorrhages appeared due to the induction of cavitation. Based on these encouraging clinical results, more extensive clinical studies have been initiated. Transcranial MRgFUS is a fast-growing field of neurological research with high clinical potential. © 2015 S. Karger AG, Basel.


Schwenke M.,Fraunhofer Institute for Medical Image Computing
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention | Year: 2011

In this paper, anisotropic Fast Marching is employed to compute blood flow trajectories as minimal paths in 3D phase-contrast MRI images. Uncertainty in the estimated blood flow vectors is incorporated in a tensor which is used as metric for the anisotropic Fast Marching. A flow connectivity distribution is computed simultaneously to the Fast Marching. Based on the connectivity distribution the most likely flow trajectories can be identified. Results are presented for several PC MRI data sets and the capability of the method to indicate uncertainty of the flow trajectories is shown.


Jacobs C.,Fraunhofer Institute for Medical Image Computing
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention | Year: 2011

Ground glass nodules (GGNs) occur less frequent in computed tomography (CT) scans than solid nodules but have a much higher chance of being malignant. Accurate detection of these nodules is therefore highly important. A complete system for computer-aided detection of GGNs is presented consisting of initial segmentation steps, candidate detection, feature extraction and a two-stage classification process. A rich set of intensity, shape and context features is constructed to describe the appearance of GGN candidates. We apply a two-stage classification approach using a linear discriminant classifier and a GentleBoost classifier to efficiently classify candidate regions. The system is trained and independently tested on 140 scans that contained one or more GGNs from around 10,000 scans obtained in a lung cancer screening trial. The system shows a high sensitivity of 73% at only one false positive per scan.

Loading Fraunhofer Institute for Medical Image Computing collaborators
Loading Fraunhofer Institute for Medical Image Computing collaborators