5th Dutch Bio-Medical Engineering Conference 2015
22-23 January 2015, Egmond aan Zee, The Netherlands
10:40   Imaging I
15 mins
Renske Hoeben, Els Butter, Bert Verdonk, Eleonora van Dorst, Herke Jan Noordmans
Abstract: Lymphedema or accumulation of lymph fluid is a debilitating complication after cancer surgery. An estimated 6-40% of patients develop lymphedema in their pelvic area and legs after gynaecologic cancer treatment [1]. To effectively treat lymphedema, it is necessary to recognize and diagnose its occurrence as early as possible. However, a scientifically proven standard to diagnose and evaluate lymphedema is currently unavailable. The aim of this study is to propose a measurement standard for the evaluation and diagnosis of lymphedema after gynaecological cancer. We performed a literature study on measurement methods for lymphedema and determined their suitability for measurements in the lower limbs and pelvic area. Many studies have reviewed the measurement quality (reproducibility and reliability) in order to determine a gold standard. It is debatable if there is a single optimal measurement technique and if a one size fits all method is required. We found the need for lymphedema measurements to be evident at four moments in the patients care cycle. We therefore propose the use of a triage model, where different measurement methods [2-4] are recommended according to the aim of the care cycle step. 1) Screening (identifying the possibility/risk of lymphedema) A general practitioner can monitor the patients general wellbeing post-surgery, and screen for the occurrence of lymphedema. Advised method(s): questionnaires, bioimpedance spectroscopy, thermography 2) Diagnosing lymphedema A physician needs information on both volume and tissue consistency to accurately diagnose lymphedema. Advised method(s): MRI 3) Treating Lymphedema (monitor treatment efficacy) A physiotherapist needs to monitor treatment efficacy. Advised method(s): circumference measurement, thermography 4) Managing lymphedema A Patient needs feedback and support to adhere to lifestyle changes, thus managing his lymphedema (and preventing progression) Advised method(s): bioimpedance spectroscopy This triage model is a first proposal to start improving early recognition of symptoms and diagnosis of lymphedema in gynaecological cancer survivors, thus preventing long term morbidity. Instead of searching for a single gold standard in lymphedema measurements we should focus on using and developing measurement methods that aid in screening, diagnosis, treatment or management of lymphedema.
15 mins
C. Beijst, M. Elschot, R.L.P. van Veen, M.A. Viergever, H.W.A.M. de Jong
Abstract: Background: The availability of real time molecular information in the intervention room, in concert with simultaneous real time anatomical information, has the potential to benefit image guided (oncological) procedures such as internal radiotherapy, tumor resections and biopsies. Currently available hybrid imaging systems, such as the SPECT/CT and the PET/CT, are impractical for interventional use due to their geometry and lack of real time and simultaneous imaging capabilities1. We present a hybrid imaging device that combines simultaneous real time x-ray and nuclear imaging of the same field of view. The goal of this study is to investigate the technical feasibility of hybrid simultaneous x-ray and nuclear imaging. Materials and Methods: Necessary components to combined fluoroscopic/nuclear imaging are an x-ray tube, an x-ray detector and a gamma camera with collimator. Our concept relies on placing these three elements in one line enabling imaging of the same field-of-view. Straight-forward combination of these elements would block the line of views. Inspired by how eyes see around the nose, a gamma camera collimator with pinholes that sees around the x-ray tube was developed. A prototype was built by combining an x-ray c-arm, a full field gamma camera and a four-pinhole collimator. The stereoscopic gamma camera views were converted to views that coincide with the x-ray view. Proof of concept was tested by manually moving a syringe with 25.9MBq Tc-99m -source through the combined field of view of both modalities. Results and conclusion: Nuclear -images (2 frames per second) were acquired simultaneously with x-ray images. With excellent spatial and temporal overlap of both modalities, x-ray images can be shown in grayscale and nuclear images in color overlay. A novel hybrid imaging prototype that combines simultaneous x-ray and nuclear imaging of the same field of view is proposed. Measurements with our concept device demonstrated that real time hybrid imaging in the intervention room is feasible. REFERENCES [1] Basu S, Alavi A. SPECT-CT and PET-CT in Oncology - An Overview. Current Medical Imaging Reviews. 2011;7:202-209.
15 mins
Joris Roels, Jan Aelterman, Jonas De Vylder, Hiep Luong, Saskia Lippens, Wilfried Philips, Yvan Saeys
Abstract: 3D Electron Microscopy (EM) has already proven to be useful in several fields such as Biotechnology and Healthcare, especially for ultrastructural analysis. However, images produced by 3D EM are in most cases severely degraded. These degradations arise due to a multitude of reasons, e.g. the complex electronics in the system, magnetic lens aberration, heating, motion stability, charging, etc. Although the raw, degraded images are currently used for analysis, their usefulness is limited because the degradations make visual distinction and automated analysis of biological features difficult. Recent research has shown the significant importance of noise characteristics as a prior in probabilistic image restoration algorithms such as Maximum A Posteriori (MAP) and Linear Minimal Mean Squared Error (LMMSE) [1, 2]. However, specific analysis of the noise characteristics in EM imaging is still an unexplored research domain and yet very important because of the unique imaging conditions, compared to ordinary light microscopy. In this work, we present a thorough analysis of noise, as one of the most important degradations in 3D EM imaging. We used statistical techniques to analyse characteristic noise properties such as distribution, stationarity, spatial correlation, signal and dwell-time dependency. This analysis allowed us to model EM noise as Gaussian, stationary noise. Furthermore, we found a horizontal correlation between neighbouring noise pixels due to horizontal scanning. This explains the striping effect we observe in noisy EM images. Also, a significant, linear signal dependency relation was derived, since low-intensity (dark) regions revealed low noise levels, compared to high-intensity (bright) regions. Lastly, we observed a significant dwell-time dependency, yielding high signal-to-noise ratios (SNR) at long dwell- times and vice versa. Therefore, the user is allowed to set the dwell-time parameter according to the desired SNR. Future work consists of a further analysis of additional artefacts (blur, charging, contamination, etc.) in 3D EM. Similar to [1], a Bayesian restoration algorithm will be developed, exploiting the artefact characteristics, in order to improve state-of-the-art techniques.
15 mins
Benjamin Schullcke, Sabine Krueger-Ziolek, Knut Moeller
Abstract: Regional ventilation of a volunteer was determined with two independent EIT devices running in parallel at different thoracic planes. Results indicate that simultaneous operation of two devices is possible if settings are chosen appropriately. With such a system e.g. shifting of lung tissue along the vertical axis can be visualized. The drawback of available EIT systems which can only visualize regional ventilation for a single electrode plane could be overcome. The described system has the potential to increase knowledge in lung mechanics and ventilation and therefore support clinical decision making.
15 mins
Vera Lagerburg, Cindy Maandag, John Wondergem
Abstract: Introduction: Diffusion Tensor Imaging (DTI) is a quite new application which necessitates protocol optimisation and testing. MRI phantoms can be used for quality control and for protocol optimisation in a standard, controlled surrounding. However currently no commercial DTI phantoms are available. The goal of this research is to develop and test a phantom for DTI which represents a human spinal cord and has ADC (Apparent Diffusion Coefficient) and FA (Fractional Anisotropy) values comparable to values found in (healthy) humans. Methods: The phantom consists of four main parts: a shield with a cover (nylon PA-I2, rapid prototyping), a tube support (polycarbonate), a tube (PTFE (polytetrafluoroethylene), (Ø 10 mm) and the fibres (polythylene terephthalate, diameter 23m). The tube with the (solid) fibres represents the human spinal cord. The fibres are pulled through the PTFE tube using a thicker wire. The phantom is filled with a Gadolinium contrast solution (Gadoteric acid, Dotarem® Guerbet). Measurements were performed on a Philips 3T MRI scanner. A standard, clinical, protocol (main parameters: Sense-NV-8 coil, ACQ voxel 3/3/2 mm, REC voxel 1.3/1.3/2 mm, TR 3185 act, TE 63, SE MS EPI SPAIR, DTI medium 0 + 800 ascending) was adapted to study the influence of different parameters on the FA and ADC value in five phantoms with a different amount of fibers and three volunteers (b-value, number of directions, resolution and TE). FA and ADC values were calculated using the Philips Software. For each measurement of ADC and FA values, 5 regions of interests have been made, which were averaged to enhance the data accuracy. The influence of the different sequence parameters on the FA and ADC values was compared with values measured in the three volunteers. Results: The FA values of the phantoms were lower than the values found in literature [1] and the volunteers, while the ADC values were higher. The influence of the sequence parameters on the FA and ADC values was comparable for the phantom and the volunteers except for the influence of the resolution. Discussion: The higher ADC and lower FA values of the phantom can be explained by the number of fibers and therefore the size of the gap between the fibers. By increasing the number of fibers the FA value will increase and the ADC value will decrease. The results of the influence of the different scanning parameters are promising, although the influence of the resolution has to be further investigated. Conclusion: A phantom for DTI is developed. First tests showed that the FA and ADC values of the phantom show comparable reactions on changes in the sequence parameters as in humans, but further research is necessary to be able to use the phantom for sequence optimization. REFERENCES [1] Mamata H, Jolesz FA, Maier SE, Apparent diffusion coefficient and fractional anisotropy in spinal cord: age and cervical spondylosis-related changes. J Magn Reson Imaging. 2005 Jul;22(1):38-43.
15 mins
Jurgen Hebbink, Hil Meijer, Geertjan Huiskamp, Frans Leijten, John Terry, Stephan van Gils
Abstract: Surgery is a possible treatment for refractory epilepsy. Prior to surgery, sometimes subdural grid electrodes are placed and brain activity is monitored for a few days. The resulting recordings are used to localize the seizure onset zone and to map areas related to critical functions such as speech and motor function. During the monitoring time, single pulse electrical stimulation (SPES) allows active exploration of neural connectivity and pathological tissue using characteristic features of the responses such as high frequency oscillations and delayed responses [1,2]. These responses contain early responses which might be used to infer connectivity. In [3] a phenomenological model of seizure generation was studied. This model has four nodes with diffusive coupling between nodes on a directed graph. The autonomous dynamics of each node is given by a noisy bistable oscillator [4]. The linear stability of the origin is governed by a dynamic variable varying between nodes. It was shown that removing an edge at random led to an increase in seizure rate in most cases and the improvement in other cases was only minor. We hypothesize that SPES may be used to derive a patient-specific model that may reveal how seizures are initiated and propagated and guide what tissue to treat. To explore this idea we used the model from [3] and added SPES stimulation to it. We first determine all pathological networks that have network seizures. For these networks we tested whether stimulation could be used to infer connectivity between nodes. It turns out that if the stimulation is sufficiently strong, we recover nearly all actual edges. Next we consider the removal of either a node or an edge to mimic surgery. We determine the pathological networks for which this result in a decrease of seizure rate. Most crucial is the ability to select the driver and to retain cycles in the network connectivity. Our study shows that active stimulation can uncover functional connectivity for a computational model if the stimulation is sufficiently strong. This connectivity informs whether removing edges or nodes diminishes the likelihood of seizure generation. To generate more realistic neurophysiological responses, we aim to use neural mass models and to include delays in the network. Further we will validate our approach using a network derived from early responses from SPES in epilepsy patients. REFERENCES [1] Valentín, A., M. Anderson, G. Alarcón, J. J. García Seoane, R. Selway, C. D. Binnie, and C. E. Polkey. 2002. Responses to single pulse electrical stimulation identify epileptogenesis in the human brain in vivo. Brain 125, no. 8: 1709-1718. [2] Van't Klooster, M. A., M. Zijlmans, F. S. S. Leijten, C. H. Ferrier, M. J. A. M. Van Putten, and G. J. M. Huiskamp. 2011. Time-frequency analysis of single pulse electrical stimulation to assist delineation of epileptogenic cortex. Brain 134, no. 10: 2855-2866. [3] Terry, J. R., O. Benjamin, and M. P. Richardson. 2012. Seizure generation: The role of nodes and networks. Epilepsia 53, no. 9: e166-e169. [4] Kalitzin, S. N., D. N. Velis, and F. H. Lopes da Silva. 2010. Stimulation-based anticipation and control of state transitions in the epileptic brain. Epilepsy and Behavior 17, no. 3: 310-323.