Regensburg Center of Biomedical Engineering - RCBE
Refine
Year of publication
Document Type
- conference proceeding (article) (92) (remove)
Has Fulltext
- no (92)
Is part of the Bibliography
- no (92)
Keywords
- Deep Learning (4)
- Simulation (4)
- Automatische Klassifikation (3)
- Bildverarbeitung (3)
- Diagnose (3)
- Lernprogramm (3)
- Magnetic Resonance Imaging (3)
- Osteosynthese (3)
- Bilderkennung (2)
- Bildsegmentierung (2)
Institute
- Regensburg Center of Biomedical Engineering - RCBE (92)
- Fakultät Maschinenbau (71)
- Labor Biomechanik (LBM) (63)
- Fakultät Informatik und Mathematik (20)
- Regensburg Medical Image Computing (ReMIC) (20)
- Labor Biofluidmechanik (6)
- Regensburg Center of Health Sciences and Technology - RCHST (5)
- Fakultät Sozial- und Gesundheitswissenschaften (2)
- Labor Empirische Sozialforschung (2)
- Labor Faserverbundtechnik (LFT) (2)
Begutachtungsstatus
- peer-reviewed (18)
- begutachtet (1)
We investigate contrastive learning in a multi-task learning setting classifying and segmenting early Barrett’s cancer. How can contrastive learning be applied in a domain with few classes and low inter-class and inter-sample variance, potentially enabling image retrieval or image attribution? We introduce a data sampling strategy that mines per-lesion data for positive samples and keeps a queue of the recent projections as negative samples. We propose a masking strategy for the NT-Xent loss that keeps the negative set pure and removes samples from the same lesion. We show cohesion and uniqueness improvements of the proposed method in feature space. The introduction of the auxiliary objective does not affect the performance but adds the ability to indicate similarity between lesions. Therefore, the approach could enable downstream auto-documentation tasks on homogeneous medical image data.
Surgical Smoke is generated during the cauterization of tissue with high-frequency (HF) devices and consists of 95% water vapor and 5% cellular debris. When the coagulation tweezers, which are supplied with HF voltage by the HF device, touch tissue, the electric circuit is closed, and smoke is generated by the heat. In-vivo investigations are performed during tracheotomies where surgical smoke is produced during coagulation of tissue. Furthermore, in-vitro parametric studies to investigate the particle number and size distribution and the spatial distribution of surgical smoke with laser light sheet technique are conducted. With higher power of the HF device, the particles generated are larger in size and the total number of particles generated is also higher. Adding artificial saliva to the tissue shows even higher particle counts. The study by laser light sheet also confirms this. The resulting characteristic size distribution, which may include viruses and bacterial components, confirms considering the risk arising from surgical smoke. Furthermore, the experiments will provide the database for further numerical investigations.
High Spatial Resolution Tomo-PIV of the Trachea Focussing on the Physiological Breathing Cycle
(2023)
Investigations of complex patient-specific flow in the nasopharynx requires high resolution numerical calculations validated by reliable experiments. When building the validation base and the benchmark of computational fluid dynamics, an experimental setup of the nasal airways was developed. The applied optical measurement technique of tomo-PIV supplies information on the governing flow field in three dimensions.
This paper presents tomo-PIV measurements of the highly complex patient-specific geometry of the human trachea. A computertomographic scan of a person’s head builds the basis of the experimental silicone model of the nasal airways. An optimised approach for precise refractive index matching avoids optical distortions even in highly complex non-free-of-sight 3D geometries. A linear-motor-driven pump generates breathing scenarios, based on measured breathing cycles. Adjusting of the CCD cameras‘ double-frame-rate PIV-Δt enables the detailed analysis of flow structures during different cycle phases. Merging regions of interest enables high spatial resolution acquisition of the flow field.
BACKGROUND:
Shear-induced conformational changes of von Willebrand factor (vWF) may be responsible for coagulation disorder and clot formation inside membrane oxygenators (MOs) during extracorporeal membrane oxygenation (ECMO) therapy.
OBJECTIVE:
The aim was to identify vWF structures inside clinically used MOs and employ computational fluid dynamics to verify the corresponding flow conditions.
METHODS:
Samples from gas exchange membranes (GEM) from MOs were analysed for accumulations of vWF and P-selectin-positive platelets using immunofluorescence techniques. Streamlines and shear rates of the flow around GEMs were computed using a laminar steady Reynolds-Averaged-Navier-Stokes approach.
RESULTS:
Most samples were colonized with equally distributed leukocytes, integrated in thin cobweb-like vWF-structures. Only 25 % of the samples showed extended accumulations of vWF. Computed streamlines showed considerable cross flow between interconnected neighbouring channels. Stagnation points were non-symmetric and contact faces were washed around closely. The occurring maximum shear rates ranged from 2,500 to 3,000 1/s.
CONCLUSIONS:
If pronounced vWF structures are present, shape and extent match the flow computations well. Computed shear rates bear a critical degree of uncertainty due to the improper viscosity model. If flow conditions inside the MO were sufficient to affect vWF, a more consistent distribution of vWF across the samples should be present.
Convolutional Neural Networks for Approximation of Internal Non-Newtonian Multiphase Flow Fields
(2021)
Neural networks (NNs) as an alternative method for universal approximation of differential equations have proven to be computationally efficient and still sufficiently accurate compared to established methods such as the finite volume method (FVM). Additionally, analysing weights and biases can give insights into the underlying physical laws. FVM and NNs are both based upon spacial discretisation. Since a Cartesian and equidistant grid is a raster graphics, image-to-image regression techniques can be used to predict phase velocity fields as well as particle and pressure distributions from simple mass flow boundary conditions. The impact of convolution layer depth and number of channels of a ConvolutionDeconvolution Regression Network (CDRN), on prediction performance of internal non-Newtownian multiphase flows is investigated. Parametric training data with 2055 sets is computed using FVM. To capture significant non-Newtownian effects of a particle-laden fluid (e.g. blood) flowing through small and non-straight channels, an Euler-Euler multiphase approach is used. The FVM results are normalized and mapped onto an equidistant grid as supervised learning target. The investigated NNs consist of n= {3, 5, 7} corresponding encoding/decoding blocks and different skip connections. Regardless of the convolution depth (i.e. number of blocks), the deepest spacial down-sampling via strided convolution is adjusted to result in a 1 × 1 × f · 2nfeature map, with f = {8, 16, 32}. The prediction performance expressed is as channel-averaged normalized root mean squared error (NRMSE). With a NRMSE of < 2 · 10-3, the best preforming NN has f = 32 initial feature maps, a kernel size of k = 4, n = 5 blocks and dense skip connections. Average inference time from this NN takes < 7 · 10-3s. Worst accuracy at NRMSE of approx 9 · 10-3is achieved without any skips, at k = 2, f = 16 and n = 3, but deployment takes only < 2 · 10-3s Given an adequate training, the prediction accuracy improves with convolution depth, where more features have higher impact on deeper NNs. Due to skip connections and batch normalisation, training is similarly efficient, regardless of the depth. This is further improved by blocks with dense connections, but at the price of a drastically larger model. Depending on geometrical complexity, spacial resolution is critical, as it increases the number of learnables and memory requirements massively.
Celiac disease is an autoimmune disorder caused by gluten that results in an inflammatory response of the small intestine.We investigated whether celiac disease can be detected using endoscopic images through a deep learning approach. The results show that additional clinical parameters can improve the classification accuracy. In this work, we distinguished between healthy tissue and Marsh III, according to the Marsh score system. We first trained a baseline network to classify endoscopic images of the small bowel into these two classes and then augmented the approach with a multimodality component that took the antibody status into account.
Vascular malformations (VMs) are a rare condition. They can be categorized into high-flow and low-flow VMs, which is a challenging task for radiologists. In this work, a very heterogeneous set of MRI images with only rough annotations are used for classification with a convolutional neural network. The main focus is to describe the challenging data set and strategies to deal with such data in terms of preprocessing, annotation usage and choice of the network architecture. We achieved a classification result of 89.47 % F1-score with a 3D ResNet 18.
Blood flow in channels of varying diameters <500μm exhibits strong non-linear effects. Multiphase finite volume approaches are feasible, but still computationally costly. Here, the feasibility of applying convolutional neural networks for blood flow prediction in artificial lungs is investigated. Training targets are precomputed using an Eulerian two-phase approach. To match with experimental data, the interphase drag and lift, as well as intraphase shear-thinning are adapted. A recursively branching regression network and convolution/deconvolution networks with plain skip connections and densely connected skips are investigated. A priori knowledge is incorporated in the loss functional to prevent the network from learning non-physical solutions. Inference from neural networks is approximately six orders of magnitude faster than the classical finite volume approach. Even if resulting in comparably coarse flow fields, the neural network predictions can be used as close to convergence initial solutions greatly accelerating classical flow computations.
Ergonomic workplaces lead to fewer work-related musculoskeletal disorders and thus fewer sick days. There are various guidelines to help avoid harmful situations. However, these recommendations are often rather crude and often neglect the complex interaction of biomechanical loading and psychological stress. This study investigates whether machine learning algorithms can be used to predict mechanical and stress-related muscle activity for a standardized motion. For this purpose, experimental data were collected for trunk movement with and without additional psychological stress. Two different algorithms (XGBoost and TensorFlow) were used to model the experimental data. XGBoost in particular predicted the results very well. By combining it with musculoskeletal models, the method shown here can be used for workplace analysis but also for the development of real-time feedback systems in real workplace environments.
BACKGROUND:
Validation and verification of multibody musculoskeletal models sEMG is a difficult process because of the reliability of sEMG data and the complex relationship of muscle force and sEMG.
OBJECTIVE:
This work aims at comparing experimentally recorded and simulated muscle activities considering a numerical model for crosstalk. METHODS:
For providing an experimentally derived reference data set, subjects were performing elevations of the arm, where the activities of the contemplated muscle groups were measured by sEMG sensors. Computed muscle activities were further processed and transformed into an artificial electromyographical signal, which includes a numerical crosstalk model. In order to determine whether the crosstalk model provides a better agreement with the measured muscle activities, the Pearson correlation coefficient has been computed as a qualitative way of assessing the curve progression of the data sets.
RESULTS:
The results show an improvement in the correlation coefficient between the experimental data and the simulated muscle activities when taking crosstalk into account.
CONCLUSIONS:
Although the correlation coefficient increased when the crosstalk model was utilized, it is questionable if the discretization of both, the crosstalk and the musculoskeletal model, is accurate enough.
High quality semi-finished products with reproducible properties of basaltic fibres are hardly available on the market. In order to examine the properties of basalt fibre reinforced plastics high quality specimens are necessary. The fibre volume content and the porosity of the produced specimen are used as the typical criteria to evaluate the achieved quality of the produced material. The effect of the geometrical properties of the different semi-finished products on the fibre volume contents is relatively small whereas the porosity content is considerably affected.
Fragestellung
Es soll in dieser Computersimulationsstudie untersucht werden, wie der Osteosyntheseverbund Platte mit Schrauben im Verbund mit einer im mittleren Drittel gebrochenen Clavicula durch das Bewegungsausmaß
in vivo belastet ist. Was sind die grundlegenden Kräfte die auf Clavicula und Implantat wirken und welchen Einfl uss hat die Bruchform.
Methodik
Die Muskel- und Gelenkkräfte sowie die Belastung des Implantatverbundes wurden mit einer muskuloskelletalen
Simulationssoftware (AnyBody Technology, V.4) berechnet. Hierfür wurden mit einem komplexen Model des menschlichen Körpers folgende Bewegungen analysiert: eine Flexion von 160° und Abduktion 160° mit
einem Gewicht von 2 kg in der Hand. Aus CT-Patientendaten wurden zwei dreidimensionale Modelle des Clavicula-Implantat Verbundes gebildet, die sich in der Frakturform unterscheiden (Querfraktur und vertikale
Fraktur). In beiden Modellen wurde eine Claviculaosteosynthese in superiorer Position mit einer 6 Loch LCP mit 2 Schrauben pro Hauptfragment verwendet. Die Materialeigenschaften wurden aus der Dichte des Materials sowie aus Literaturdaten verwendet. Die Muskel- und Gelenkkräfte aus der muskuloskelletalen Berechnung wurden auf das Finite Elemente Modell übertragen und die Spannungen und Dehnungen des Implantat-Knochenverbundes
wurden berechnet.
Ergebnisse
Es zeigte sich, dass die simulierte in vivo Belastung stark abhängig vom Flexionswinkel ist. Das Implantat ist in der superioren Lage auf Biegung belastet, welche maximale Werte im Überschulterniveau erreicht. Die Bruchform mit anatomischer Reposition und Kontakt der Hauptfragmente zueinander führt zu einer deutlichen Entlastung des Osteosyntheseverbundes im Vergleich zu einer Bruchform mit vertikaler Fraktur.
Schlussfolgerung
Aus den Analysen ist eine Positionierung der Plattenosteosynthese für die im mittleren Drittel frakturierte Clavicula in anterior-superiorer Lage wünschenswert. Die anatomische Reposition entlastet den Osteosyntheseverbund und sollte möglichst erreicht werden. Die Nachbehandlung sollte ein Bewegungsausmaß für den Arm für 4 Wochen für einfache Bruchformen auf 70° Flexion und Abduktion limitieren und für komplexe Bruchformen diese Limitierung ausgedehnt werden.
Subject-specific Musculoskeletal Simulation of Hip Dislocation Risk in Activities of Daily Living
(2011)
On modelling spine curvature dependent on muscular and external forces in multibody dynamics system
(2011)
This paper presents a computational approach for investigating effect of muscular and external forces on curvature of the lumbar spine. Multibody dynamics system is used to compute the lumbar spine curvature using a force-dependent kinematics facility, e.g. this method allows releasing some degrees of freedom in order to be computed based on the current load configuration.
INTRODUCTION
Several clinical studies demonstrated long-term adjacent-level effects after implantation of spinal fusion devices[1]. These effects have been reported as adjacent joint degeneration and the development of new symptoms correlating with adjacent segment degeneration[2] and the trend has therefore gone to motion preservation devices; however, these effects have not been understood very well and have not been investigated thoroughly[3].
The aim of this study is to investigate the effect of varying the stiffness of spinal fusion devices on the adjacent vertebral levels. Disc forces, moments and facet joint forces were analyzed.
METHODS
The AnyBody Modeling System was used to compute the in-vivo muscle and joint reaction forces of a musculoskeletal model. The full body model used in this study consists of 188 muscle fascicles in the lumbar spine and more than 1000 individual muscle branches in total. The model has been proposed by de Zee et al.[3], validated by Rasmussen et al.[4] and by Galibarov et al.[5]. The new model[5] determines the individual motions between vertebrae based on the equilibrium between forces acting on the vertebrae from muscles and joints and the passive stiffness in disks and ligaments, figure 1a. An adult of 1.75 m and 75 kg with a spinal implant in L4L5 was modeled. This model was subjected to a flexion-extension motion using different elastic moduli to analyze and compare to a non-implanted scenario. The analyzed variables were vertebral motion, the disc reaction forces and moments, as well as facet joint forces in the treated and the adjacent levels: L2L3, L3L4, L4L5 and L5-Sacrum.
RESULTS
When introducing a spinal fusion device in the L4L5 joint the reaction forces and moments decreased in this joint with stiffer devices leading to lower joint loads. However, in the adjacent joints, L3L4 and L5Sacrum, an increase was observed when implanting stiffer devices. Similar trends could be found for the L2L3 joint. The loads in the facet joints showed the same trends. While introducing a spinal fusion device reduced the facet joint forces in the treated joint, the loads in the adjacent facet joints were increased according to the stiffness of the implanted device, figure 1b.
DISCUSSION
While the treated disc joint showed reduced motion and loads, the adjacent levels demonstrated a significant increase. In particular, the increased facet joint forces in the adjacent levels can lead to adjacent level facet pain or accelerated facet joint degeneration. Introducing a device resulted in preventing facet contact and therefore facet joint loads, even using the device with the lowest stiffness.
CONCLUSION
The presented model shows that clinical complications such as facet joint degeneration in adjacent levels after implantation of spinal fusion device are consistent with the change in the mechanical-stimulus distribution in the system.
Einführung in die Mechanik
(2013)
High loads can appear in the individual joints of the human foot while the driver uses the pedals, in particular, during breaking. Measuring these internal forces is very difficult or almost impossible; therefore, advanced models are necessary to perform musculoskeletal simulations. The objective of this investigation was to see what loads are acting in the individual foot joints from the phalanges to calcaneus and talus during different brake scenarios. The Glasgow-Maastricht AnyBody Foot Model with 26 separate segments, connected by joints, ligaments and muscles was used inside the AnyBody Modeling System to compute individual mid foot joint loads. The amount, the direction of the force and additionally also the load insertion point was varied for several simulations. Figure 1: Seated musculoskeletal body model with applied brake force and forces for the lateral, intermediate and medial cuneiform-navicular joint for two different brake forces. The simulation showed that for the different brake scenarios, different muscles will be activated in the human and therefore different loads are apply in the fore-and mid-foot, respectively. The torso of the subject was assumed to be fixed in the seat. Further studies are ongoing to simulate the seat as an elastic element that allows different H-point locations according to the different loadings in the foot from the brake pedal using a new inverse dynamics analysis method called force-dependent kinematics.
Navigated Femur First Total Hip Arthroplasty leads to improved Biomechanical Outcome after surgery
(2015)
Thoracolumbar spine model with articulated rigcage for the prediction of dynamic spinal loading
(2015)
The musculoskeletal load scenario of computer-assisted Femur-First THR up to one year after surgery
(2015)
Biomechanik des Alterns
(2016)
BACKGROUND: Validation and verification of multibody musculoskeletal models sEMG is a difficult process because of the reliability of sEMG data and the complex relationship of muscle force and sEMG. OBJECTIVE: This work aims at comparing experimentally recorded and simulated muscle activities considering a numerical model for crosstalk. METHODS: For providing an experimentally derived reference data set, subjects were performing elevations of the arm, where the activities of the contemplated muscle groups were measured by sEMG sensors. Computed muscle activities were further processed and transformed into an artificial electromyographical signal, which includes a numerical crosstalk model. In order to determine whether the crosstalk model provides a better agreement with the measured muscle activities, the Pearson correlation coefficient has been computed as a qualitative way of assessing the curve progression of the data sets. RESULTS: The results show an improvement in the correlation coefficient between the experimental data and the simulated muscle activities when taking crosstalk into account. CONCLUSIONS: Although the correlation coefficient increased when the crosstalk model was utilized, it is questionable if the discretization of both, the crosstalk and the musculoskeletal model, is accurate enough.
In recent years, musculoskeletal computation has become a widely used tool to investigate joint and muscle forces within the human body. However, the issue of muscle fatigue is not considered adequately in most models and is a challenging task. One aspect that needs to be examined is the interaction of muscles during an exhausting task. Therefore, an experimental study was designed to analyze the changes of back muscle recruitment pattern during such exercises.
In this study 38 subjects (27 male, 11 female, height = 177±8.5 cm, weight = 74.0±13.6 kg) participated. Each subject had to perform three static and three dynamic exhausting exercises where the back muscles were loaded with subject specific forces using a dynamometer adapter especially designed for the trunk muscles. To collect the muscle activity, twelve surface electromyography sensors were applied on the back, and four on the abdominal muscles. Muscle activity and fatigue were analyzed by calculating the maximum voluntary contraction normalized signal and the median frequency. At first the fatigue of m. erector spinae and m. multifidi was analyzed, since these muscles carry the main load during the exercises. Subsequently the activity of the m. trapezius, m. rectus abdominis and m. obliquus externus were investigated to determine recruitment patterns. To gain more detailed information of these patterns a numerical model was built using the AnyBody Modeling System™. Analyzing the measurements, we can observe an increasing muscle activity during isokinetic exercises while the force is constant. Since the activity in the simulation is defined as the current force output divided by the strength of the muscle, the strength parameter was scaled down based on the measured data, assuming a linear force – activity correlation, and using a numerical algorithm considering the influence of cross talk.
The results show, that changes in recruitment pattern can be divided into three major subgroups. Prior to total exhaustion, some of the subjects show additional activation of muscles in the trapezius region, while other subjects show an additional activation of abdominal muscles, increasing the intra-abdominal pressure which supports the spine. In the third group an activation in both regions can be observed.
The numerical simulations show an increasing activity of abdominal muscles as well as muscles in the upper back. Especially the m. latissimus dorsi shows a significantly higher activity.
The results lead to the conclusion that prior to total exhaustion, additional muscles are recruited to support the main muscles. It was shown that abdominal muscles are activated to support back muscles by pressurizing the trunk cavity to delay total exhaustion as long as possible. In conclusion, the results show that changes in muscle recruitment pattern need to be considered when introducing muscle fatigue to musculoskeletal models.
The AnyBody™ Modeling System (AMS) [1], is an universally used musculoskeletal simulation software using inverse dynamics. Until now, no complete human hand model is known in the AMS. Also considering other musculoskeletal software platforms, just one detailed entire hand model is recently published [2] but is only based on one subject. The aim of this work is to implement a full detailed hand model for the AMS including all extrinsic and intrinsic muscles using data by the UWB gained through an anatomical study of ten cadaver hands.
Soccer is one of the most popular sports all around the world. It is an injurious type of sport with a focus on lower extremities and high psychological pressure during matches. The stressor is linked with injuries and an increased musculoskeletal loading. This study investigates the influence of cognitive stress on the load profile of the knee joint. Twelve professional youth soccer players performed highly dynamic runs with and without additional cognitive stress. The runs were analysed with a musculoskeletal simulation software. The data analysis shows no difference in knee joint reaction loading under additional mental stress compared to the baseline. Yet running times are significantly lower in the baseline. While there is no increase in the joint loads, the running times indicate an altered movement behaviour when the subjects are exposed to additional mental demand.
The gold standard for the analysis of human kinematics and kinetics is a camera-based motion capture system in combination with force measurement platforms. Alternatively, inertial measurement units can be utilized to obtain human kinematics, while ground reaction forces are computed from full body dynamics. This setup represents a system independent from the spatial confinement of a gait laboratory. The aim of this study is the comparison of the two methods by the investigation of lower limb kinematics and the resulting joint reaction forces within the ankle-, knee- and hip joints. For this purpose, human motion during gait was captured simultaneously by both measurement techniques. 13 trials from 8 different test subjects were evaluated in total. IMU data was processed with a quaternion based Kalman Filter. The data sets were implemented into a musculoskeletal simulation program in order to drive a virtual human body model. Each sensor was aligned to the gravitational and magnetic field vectors of the earth. The angles of flexions, extensions and rotations were analyzed to determine kinematic differences. Joint reaction forces defined kinetic dissimilarities. The overall kinematic differences of both models yielded root mean square errors of 7.62°, 6.02°, 4.95°, 2.79°, 2.38° and 3.56° for ankle flexion, subtalar eversion, knee flexion, hip external rotation, hip abduction and hip flexion, respectively. The proximo-distal differences in force peaks between the models yielded overall for the ankle, 57.33 %Bodyweight(BW) ± 46.86 %BW (16.66 %(Maximum peak to peak) ± 13.62 %) for the knee 37.09 %BW ± 29.33 %BW (17.65 % ± 15.44 %) and 32.03 %BW ± 24.33 %BW (15.6 % ± 12.54 %) for the hip. The overall outcome of this work investigated an approach independent of the common setup of the gait laboratory, thus enabling a cheaper and more flexible technology as an alternative. However, kinematic and thus kinetic differences remain rather large. Future work aims to improve the contact criterion for the calculation of the ground reaction forces and the implementation of a full-body calibration algorithm for the IMU system in order to counteract magnetic field disturbances.
Common Schanz screw systems can be used to stabilize pelvic ring fractures. In order to accommodate for different patient’s requirements, implants can be placed in cranio-caudal direction into the os ilium (T1), or into the supraacetabular bone canal, and thus, in dorso-ventral direction (T2). Whereas both techniques are currently used, no data of the biomechanical behavior is available up to this date. The aim of this study is to analyze, whether T2 shows biomechanical advantages with respect to tissue and implant stresses due to the enlarged bone-implant interface. Forces acting on the pelvis were analyzed using motion capture data of a gait cycle obtained by the utilization of a musculoskeletal simulation program. A three dimensional finite element (FE) model of the pelvis with grayscale-based material properties was generated. The muscle and joint reaction forces at toe-off were applied to the FE model and instable pelvis fractures were implemented. The osteosynthesis systems were positioned within the model in order to enable the comparison between the two different surgical techniques. Stresses and displacements were analyzed for bone tissue, fracture zone and implant.
T2 lead to approx. 30% larger displacements in the fracture zone. Von-Mises stresses were larger for T2 in the implant (80 MPa vs. 227 MPa), whereas T1 leads to larger stresses in the bone tissue (200 MPa vs. 140 MPa).
Both implantation techniques showed a good biomechanical behavior. Differences could be found with respect to tissue strains and deformations in the fracture zone. If bone quality or fracture healing are of concern, T2 or T1 should be used, respectively. However, both techniques seem to be applicable for cases with no special requirements. Further analyses aim to investigate the behavior under cyclic loading.
Simulation of Fluid-Structure Interaction between injection medium and balloon catheter using ICFD
(2017)
Arteriosclerosis is a major health issue worldwide. While it is commonly treated by the implantation of an balloon-expandable stent, micro injuries may occur during stent deployment, and induce in-stent restenosis, whose consequence can be fatal. Studying this undesirable phenomenon is usually limited as experimental data is hard to obtain on ethical ground. Numerical simulation are performed to better understand this problem. To construct a more realistic simulation of a balloon-expandable stent, a partitioned strongly-coupled FSI simulation of the balloon deployment was set up using the ICFD solver of LS-DYNA, - a quite innovative approach. The complex balloon configuration as well as the interaction of the injection medium and the balloon structure was considered. The balloon structure consisting of shell elements was obtained from preliminary balloon folding and pleating simulations. The balloon consists of a flexible thin walled polyamide. The injection fluid is implemented using volume elements. Balloon deployment was initiated by a pressure boundary condition inducing a volume flow into the balloon. The initial feasibility analysis showed promising result including a continuous balloon deployment and a reasonable development of the fluid pressure and velocity field. However, applying this FSI approach to a more complex balloon structure led to a non convergent solution. The non-convergence could be mainly reduced to mechanical factors including the low wall thickness of the balloon (< 0.05 mm) and the flexibility of the polyamide. Further, the ICFD solver shows less accuracy concerning the FSI conditions when dealing with thin flexible structures as well as enclosed volumes. A shell thickness of 0.06 mm is believed to result in a convergent solution.
The development of biomimetic robots has gained research interest in the last years as it may both help under-standing processes of motion execution in biological systems as well as developping a novel generation of intelligent and energy efficient robots. However, exact model generation that builds up on observations and robot design is very time intensive. In this paper we present a novel pipeline for co-development of biomimetic hardware and simulation models based on biological Computer Tomography (CT) data. For this purpose we exploit State of the Art rapid prototyping technologies such as 3D Printing and the Neurorobotics Platform for musculoskeletal simulations in virtual environments. The co-development integrates both advantages of virtual and physical experimental models and is expected to increase development speed of controllers that can be tested on the simulated counterpart before application to a printed robot model. We demonstrate the pipeline by generating a one year old infant model as a musculoskeletal simulation model and a print-in-place 3D printed skeleton as a single movable part. Even though we hereonly introduce the initial body generation and only a first testsetup for a modular sensory and control framework, we can clearly spot advantages in terms of rapid model generation and highly biological related models. Engineering costs are reducedand models can be provided to a wide research community for controller testing in an early development phase.
In this paper, we address the problem of retrospective color shading correction. An extension of the established gray-level shading correction algorithm based on signal envelope (SE) estimation to color images is developed using principal color components. Compared to the probably most general shading correction algorithm based on entropy minimization, SE estimation does not need any computationally expensive optimization and thus can be implemented more effciently. We tested our new shading correction scheme on artificial as well as real endoscopic images and observed promising results. Additionally, an indepth analysis of the stop criterion used in the SE estimation algorithm is provided leading to the conclusion that a fixed, user-defined threshold is generally not feasible. Thus, we present new ideas how to develop a non-parametric version of the SE estimation algorithm using entropy.
Chest X-Ray (CXR) images as part of a non-invasive diagnosis method are commonly used in today’s medical workflow. In traditional methods, physicians usually use their experience to interpret CXR images, however, there is a large interobserver variance. Computer vision may be used as a standard for assisted diagnosis. In this study, we applied an encoder-decoder neural network architecture for automatic lung region detection. We compared a three-class approach (left lung, right lung, background) and a two-class approach (lung, background). The differentiation of left and right lungs as direct result of a semantic segmentation on basis of neural nets rather than post-processing a lung-background segmentation is done here for the first time. Our evaluation was done on the NIH Chest X-ray dataset, from which 1736 images were extracted and manually annotated. We achieved 94:9% mIoU and 92% mIoU as segmentation quality measures for the two-class-model and the three-class-model, respectively. This result is very promising for the segmentation of lung regions having the simultaneous classification of left and right lung in mind.
Work places and conditions strains the human body, both psychologically and biomechanically. In order to analyse working conditions and in the following to improve them, detailed knowledge about the effect of the different stressors on the body is needed. This manuscript discusses methods on how to evaluate biomechanical and mental loading and its effect on the musculoskeletal system. A possible workflow for the analysis is presented.
Serial 2D section images with high resolution, resulting from innovative imaging methods become even more valuable, if they are fused with in vivo volumes. Achieving this goal, the 3D context of the sections would be restored, the deformations would be corrected and the artefacts would be eliminated. However, the registration in this field faces big challenges and is not solved in general. On the other hand, several approaches have been introduced dealing at least with some of these difficulties. Here, a brief overview of the topic is given and some of the solutions are presented. It does not constitute the claim to be a complete review, but could be a starting point for those who are interested in this field.
GraphMIC is a cross-platform image processing application utilizing the libraries ITK and OpenCV. The abstract structure of image processing pipelines is visually represented by user interface components based on modern QtQuick technology and allows users to focus on arrangement and parameterization of operations rather than implementing the equivalent functionality natively in C++. The application's central goal is to improve and simplify the typical workflow by providing various high level features and functions like multi threading, image sequence processing and advanced error handling. A built-in python interpreter allows the creation of custom nodes, where user defined algorithms can be integrated to extend basic functionality. An embedded 2d/3d visual-izer gives feedback of the resulting image of an operation or the whole pipeline. User inputs like seed points, contours or regions are forwarded to the processing pipeline as parameters to offer semi-automatic image computing. We report the main concept of the application and introduce several features and their implementation. Finally, the current state of development as well as future perspectives of GraphMIC are discussed
The development of adenocarcinoma in Barrett’s esophagus is difficult to detect by endoscopic surveillance of patients with signs of dysplasia. Computer assisted diagnosis of endoscopic images (CAD) could therefore be most helpful in the demarcation and classification of neoplastic lesions. In this study we tested the feasibility of a CAD method based on Speeded up Robust Feature Detection (SURF). A given database containing 100 images from 39 patients served as benchmark for feature based classification models. Half of the images had previously been diagnosed by five clinical experts as being ”cancerous”, the other half as ”non-cancerous”. Cancerous image regions had been visibly delineated (masked) by the clinicians. SURF features acquired from full images as well as from masked areas were utilized for the supervised training and testing of an SVM classifier. The predictive accuracy of the developed CAD system is illustrated by sensitivity and specificity values. The results based on full image matching where 0.78 (sensitivity) and 0.82 (specificity) were achieved, while the masked region approach generated results of 0.90 and 0.95, respectively.
The Impact of Semi-Automated Segmentation and 3D Analysis on Testing New Osteosynthesis Material
(2017)
A new protocol for testing osteosynthesis material postoperatively combining semi-automated segmentation and 3D analysis of surface meshes is proposed. By various steps of transformation and measuring, objective data can be collected. In this study the specifications of a locking plate used for mediocarpal arthrodesis of the wrist were examined. The results show, that union of the lunate, triquetrum, hamate and capitate was achieved and that the plate is comparable to coexisting arthrodesis systems. Additionally, it was shown, that the complications detected correlate to the clinical outcome. In synopsis, this protocol is considered beneficial and should be taken into account in further studies.
Computer-assisted analysis of endoscopic images can be helpful to the automatic diagnosis and classification of neoplastic lesions. Barrett's esophagus (BE) is a common type of reflux that is not straight forward to be detected by endoscopic surveillance, thus being way susceptible to erroneous diagnosis, which can cause cancer when not treated properly. In this work, we introduce the Optimum-Path Forest (OPF) classifier to the task of automatic identification of Barrett'sesophagus, with promising results and outperforming the well known Support Vector Machines (SVM) in the aforementioned context. We consider describing endoscopic images by means of feature extractors based on key point information, such as the Speeded up Robust Features (SURF) and Scale-Invariant Feature Transform (SIFT), for further designing a bag-of-visual-wordsthat is used to feed both OPF and SVM classifiers. The best results were obtained by means of the OPF classifier for both feature extractors, with values lying on 0.732 (SURF) - 0.735(SIFT) for sensitivity, 0.782 (SURF) - 0.806 (SIFT) for specificity, and 0.738 (SURF) - 0.732 (SIFT) for the accuracy.
Die Implantation eines Cochlea Implantates benötigt einen chirurgischen Zugang im Felsenbein und durch die Paukenhöhle des Patienten. Der Chirurg hat eine eingeschränkte Sicht im Operationsgebiet, die weiterhin viele Risikostrukturen enthält. Um eine Cochlea Implantation sicher und fehlerfrei durchzuführen, ist eine umfangreiche theoretische und praktische (teilweise berufsbegleitende) Fortbildung sowie langjährige Erfahrung notwendig. Unter Nutzung von realen klinischen CT/MRT Daten von Innen- und Mittelohr und der interaktiven Segmentierung der darin abgebildeten Strukturen (Nerven, Cochlea, Gehörknöchelchen,...) wird im HaptiVisT Projekt ein haptisch-visuelles Trainingssystem für die Implantation von Innen- und Mittelohr-Implantaten realisiert, das als sog. „Serious Game“ mit immersiver Didaktik gestaltet wird. Die Evaluierung des Demonstrators hinsichtlich Zweckmäßigkeit erfolgt prozessbegleitend und ergebnisorientiert, um mögliche technische oder didaktische Fehler vor Fertigstellung des Systems aufzudecken. Drei zeitlich versetzte Evaluationen fokussieren dabei chirurgisch-fachliche, didaktische sowie haptisch-ergonomische Akzeptanzkriterien.
Introduction – Diagnosis of abnormal cilia function is based on ultrastructural analysis of axoneme defects, especialy the features of inner and outer dynein arms which are the motors of ciliar motility. Sub-optimal biopsy material, methodical, and intrinsic electron microscopy factors pose difficulty in ciliary defects evaluation. We present a computer-assisted approach based on state-of-the-art image analysis and object recognition methods yielding a time-saving and efficient diagnosis of cilia dysfunction. Method – The presented approach is based on a pipeline of basal image processing methods like smoothing, thresholding and ellipse fitting. However, integration of application specific knowledge results in robust segmentations even in cases of image artifacts. The method is build hierarchically starting with the detection of cilia within the image, followed by the detection of nine doublets within each analyzable cilium, and ending with the detection of dynein arms of each doublet. The process is concluded by a rough classification of the dynein arms as basis for a computer-assisted diagnosis. Additionally, the interaction possibilities are designed in a way, that the results are still reproducible given the completion report. Results – A qualitative evaluation showed reasonable detection results for cilia, doublets and dynein arms. However, since a ground truth is missing, the variation of the computer-assisted diagnosis should be within the subjective bias of human diagnosticians. The results of a first quantitative evaluation with five human experts and six images with 12 analyzable cilia showed, that with default parameterization 91.6% of the cilia and 98% of the doublets were found. The computer-assisted approach rated 66% of those inner and outer dynein arms correct, where all human experts agree. However, especially the quality of the dynein arm classification may be improved in future work.
Structural MRI brain analysis and segmentation is a crucial part in the daily routine in neurosurgery for intervention planning. Exemplarily, the free software FSL-FAST (FMRIB’s Segmentation Library – FMRIB’s Automated Segmentation Tool) in version 4 is used for segmentation of brain tissue types. To speed up the segmentation procedure by parallel execution, we transferred FSL-FAST to a General Purpose Graphics Processing Unit (GPGPU) using Open Computing Language (OpenCL) [1]. The necessary steps for parallelization resulted in substantially different and less useful results. Therefore, the underlying methods were revised and adapted yielding computational overhead. Nevertheless, we achieved a speed-up factor of 3.59 from CPU to GPGPU execution, as well providing similar useful or even better results.
We propose an automatic approach for early detection of adenocarcinoma in the esophagus. High-definition endoscopic images (50 cancer, 50 Barrett) are partitioned into a dataset containing approximately equal amounts of patches showing cancerous and non-cancerous regions. A deep convolutional neural network is adapted to the data using a transfer learning approach. The final classification of an image is determined by at least one patch, for which the probability being a cancer patch exceeds a given threshold. The model was evaluated with leave one patient out cross-validation. With sensitivity and specificity of 0.94 and 0.88, respectively, our findings improve recently published results on the same image data base considerably. Furthermore, the visualization of the class probabilities of each individual patch indicates, that our approach might be extensible to the segmentation domain.
Die Lehre der medizinischen Bildverarbeitung vermittelt Kenntnisse mit einem breiten Methodenspektrum. Neben den Grundlagen der Verfahren soll ein Gefühl für eine geeignete Ausführungsreihenfolge und ihrer Wirkung auf medizinische Bilddaten entwickelt werden. Die Komplexität der Methoden erfordert vertiefte Programmierkenntnisse, sodass bereits einfache Operationen mit großem Programmieraufwand verbunden sind. Die Software GraphMIC stellt Bildverarbeitungsoperationen in Form interaktiver Knoten zur Verfügung und erlaubt das Arrangieren, Parametrisieren und Ausführen komplexer Verarbeitungssequenzen in einem Graphen. Durch den Fokus auf das Design einer Pipeline, weg von sprach- und frameworkspezifischen Implementierungsdetails, lassen sich grundlegende Prinzipien der Bildverarbeitung anschaulich erlernen. In diesem Beitrag stellen wir die visuelle Programmierung mit GraphMIC der nativen Implementierung äquivalenter Funktionen gegenüber. Die in C++ entwickelte Applikation basiert auf Qt, ITK, OpenCV, VTK und MITK.
Virtual training of bone milling requires realtime and realistic haptics of the interaction between the ”virtual mill” and a ”virtual bone”. We propose an exponential abrasion model between a virtual one and the mill bit and combine it with a coarse representation of the virtual bone and the mill shaft for collision detection using the Bullet Physics Engine. We compare our exponential abrasion model to a widely used linear abrasion model and evaluate it quantitatively and qualitatively. The evaluation results show, that we can provide virtual milling in real-time, with an abrasion behavior similar to that proposed in the literature and with a realistic feeling of five different surgeons.
In order to fix a fracture using minimally invasive surgery approaches, surgeons are drilling complex and tiny bones with a 2 dimensional X-ray as single imaging modality in the operating room. Our novel haptic force-feedback and visual assisted training system will potentially help hand surgeons to learn the drilling procedure in a realistic visual environment. Within the simulation, the collision detection as well as the interaction between virtual drill, bone voxels and surfaces are important. In this work, the chai3d collision detection and force calculation algorithms are combined with a physics engine to simulate the bone drilling process. The chosen Bullet-Physics-Engine provides a stable simulation of rigid bodies, if the collision model of the drill and the tool holder is generated as a compound shape. Three haptic points are added to the K-wire tip for removing single voxels from the bone. For the drilling process three modes are proposed to emulate the different phases of drilling in restricting the movement of a haptic device.
Im Rahmen der Entwicklung eines haptisch-visuellen Trainingssystems für das Fräsen am Felsenbein werden ein Haptikarm und ein autostereoskopischer 3D-Monitor genutzt, um Chirurgen die virtuelle Manipulation von knöchernen Strukturen im Kontext eines sog. Serious Game zu ermöglichen. Unter anderem sollen Assistenzärzte im Rahmen ihrer Ausbildung das Fräsen am Felsenbein für das chirurgische Einsetzen eines Cochlea-Implantats üben können. Die Visualisierung des virtuellen Fräsens muss dafür in Echtzeit und möglichst realistisch modelliert, implementiert und evaluiert werden. Wir verwenden verschiedene Raycasting Methoden mit linearer und Nearest Neighbor Interpolation und vergleichen die visuelle Qualität und die Bildwiederholfrequenzen der Methoden. Alle verglichenen Verfahren sind sind echtzeitfähig, unterscheiden sich aber in ihrer visuellen Qualität.