We present a probabilistic framework to estimate the accumulated radiation dose and the corresponding dose uncertainty that is delivered to important anatomical structures, e.g. the primary tumor and healthy surrounding organs, during radiotherapy. The dose uncertainty we report is a direct result of uncertainties in the estimates of the deformation which aligns the daily cone-beam CT images with the planning CT. The accumulated radiation dose is an important measure to monitor during treatment, in particular to see if it significantly deviates from the planned dose which might indicate that either the patient was not properly positioned before treatment or that the anatomy has changed due to the treatment. In the case of the latter, the treatment plan should be adaptively changed to align with the current patient anatomy. We estimate the accumulated dose distribution, and its uncertainty, retrospectively on a dataset acquired during treatment of cancer in the neck and show the dose distributions in the form of dose volume histograms.
Yogesh Rathi, Oleg V Michailovich, Kawin Setsompop, Sylvain Bouix, Martha E Shenton, and Carl-Fredrik Westin. 9/2011. “Sparse Multi-shell Diffusion Imaging.” Med Image Comput Comput Assist Interv, 14, Pt 2, Pp. 58-65.Abstract
Diffusion magnetic resonance imaging (dMRI) is an important tool that allows non-invasive investigation of neural architecture of the brain. The data obtained from these in-vivo scans provides important information about the integrity and connectivity of neural fiber bundles in the brain. A multi-shell imaging (MSI) scan can be of great value in the study of several psychiatric and neurological disorders, yet its usability has been limited due to the long acquisition times required. A typical MSI scan involves acquiring a large number of gradient directions for the 2 (or more) spherical shells (several b-values), making the acquisition time significantly long for clinical application. In this work, we propose to use results from the theory of compressive sampling and determine the minimum number of gradient directions required to attain signal reconstruction similar to a traditional MSI scan. In particular, we propose a generalization of the single shell spherical ridgelets basis for sparse representation of multi shell signals. We demonstrate its efficacy on several synthetic and in-vivo data sets and perform quantitative comparisons with solid spherical harmonics based representation. Our preliminary results show that around 20-24 directions per shell are enough for robustly recovering the diffusion propagator.
We formulate registration-based elastography in a probabilistic framework and apply it to study lung elasticity in the presence of emphysematous and fibrotic tissue. The elasticity calculations are based on a Finite Element discretization of a linear elastic biomechanical model. We marginalize over the boundary conditions (deformation) of the biomechanical model to determine the posterior distribution over elasticity parameters. Image similarity is included in the likelihood, an elastic prior is included to constrain the boundary conditions, while a Markov model is used to spatially smooth the inhomogeneous elasticity. We use a Markov Chain Monte Carlo (MCMC) technique to characterize the posterior distribution over elasticity from which we extract the most probable elasticity as well as the uncertainty of this estimate. Even though registration-based lung elastography with inhomogeneous elasticity is challenging due the problem's highly underdetermined nature and the sparse image information available in lung CT, we show promising preliminary results on estimating lung elasticity contrast in the presence of emphysematous and fibrotic tissue.
In addition to functional localization and integration, the problem of determining whether the data encode some information about the mental state of the subject, and if so, how this information is represented has become an important research agenda in functional neuroimaging. Multivariate classifiers, commonly used for brain state decoding, are restricted to simple experimental paradigms with a fixed number of alternatives and are limited in their representation of the temporal dimension of the task. Moreover, they learn a mapping from the data to experimental conditions and therefore do not explain the intrinsic patterns in the data. In this paper, we present a data-driven approach to building a spatio-temporal representation of mental processes using a state-space formalism, without reference to experimental conditions. Efficient Monte Carlo algorithms for estimating the parameters of the model along with a method for model-size selection are developed. The advantages of such a model in determining the mental-state of the subject over pattern classifiers are demonstrated using an fMRI study of mental arithmetic.
Diffusion tensor magnetic resonance imaging (DTI) is a relatively new technology that is popular for imaging the white matter of the brain. This article provides a basic and broad overview of DTI to enable the reader to develop an intuitive understanding of these types of data, and an awareness of their strengths and weaknesses.
Image registration is the process of transforming images acquired at different time points, or with different imaging modalities, into the same coordinate system. It is an essential part of any neurosurgical planning and navigation system because it facilitates combining images with important complementary, structural, and functional information to improve the information based on which a surgeon makes critical decisions. Brigham and Women's Hospital (BWH) has been one of the pioneers in developing intraoperative registration methods for aligning preoperative and intraoperative images of the brain. This article presents an overview of intraoperative registration and highlights some recent developments at BWH.
User interaction is required for reliable segmentation of brain tumors in clinical practice and in clinical research. By incorporating current research tools, 3D Slicer provides a set of interactive, easy to use tools that can be efficiently used for this purpose. One of the modules of 3D Slicer is an interactive editor tool, which contains a variety of interactive segmentation effects. Use of these effects for fast and reproducible segmentation of a single glioblastoma from magnetic resonance imaging data is demonstrated. The innovation in this work lies not in the algorithm, but in the accessibility of the algorithm because of its integration into a software platform that is practical for research in a clinical setting.
Maghemite (γ-Fe(2)O(3))/multi-walled carbon nanotubes (MWCNTs) hybrid-materials were synthesized and their anisotropic electrical conductivities as a result of their alignment in a polymer matrix under an external magnetic field were investigated. The tethering of γ-Fe(2)O(3) nanoparticles on the surface of MWCNT was achieved by a modified sol-gel reaction, where sodium dodecylbenzene sulfonate (NaDDBS) was used in order to inhibit the formation of a 3D iron oxide gel. These hybrid-materials, specifically, magnetized multi-walled carbon nanotubes (m-MWCNTs) were readily aligned parallel to the direction of a magnetic field even when using a relatively weak magnetic field. The conductivity of the epoxy composites formed in this manner increased with increasing m-MWCNT mass fraction in the polymer matrix. Furthermore, the conductivities parallel to the direction of magnetic field were higher than those in the perpendicular direction, indicating that the alignment of the m-MWCNT contributed to the enhancement of the anisotropic electrical properties of the composites in the direction of alignment.
This study investigated the relationship of brain white matter (WM) lesions affecting specific neural networks with decreased mobility in ninety-nine healthy community-dwelling subjects ≥75 years old prospectively enrolled by age and mobility status. We assessed lesion burden in the genu, body and splenium of corpus callosum; anterior, superior and posterior corona radiata; anterior and posterior limbs of internal capsule; corticospinal tract; and superior longitudinal fasciculus. Burden in the splenium of corpus callosum (SCC) demonstrated the highest correlation particularly with walking speed (r=0.4, p<10(-4)), and in logistic regression it was the best regional predictor of low mobility performance. We also found that independent of mobility, corona radiata has the largest lesion burden with anterior (ACR) and posterior (PCR) aspects being the most frequently affected. The results suggest that compromised inter-hemispheric integration of visuospatial information through the SCC plays an important role in mobility impairment in the elderly. The relatively high lesion susceptibility of ACR and PCR in all subjects may obscure the importance of these lesions in mobility impairment.
Glioma histologies are the primary factor in prognostic estimates and are used in determining the proper course of treatment. Furthermore, due to the sensitivity of cranial environments, real-time tumor-cell classification and boundary detection can aid in the precision and completeness of tumor resection. A recent improvement to mass spectrometry known as desorption electrospray ionization operates in an ambient environment without the application of a preparation compound. This allows for a real-time acquisition of mass spectra during surgeries and other live operations. In this paper, we present a framework using sparse kernel machines to determine a glioma sample's histopathological subtype by analyzing its chemical composition acquired by desorption electrospray ionization mass spectrometry.
Obstructive sleep apnea (OSA) is accompanied by neurocognitive impairment, likely mediated by injury to various brain regions. We evaluated brain morphological changes in patients with OSA and their relationship to neuropsychological and oximetric data. Sixteen patients affected by moderate-severe OSA (age: 55.8±6.7 years, 13 males) and fourteen control subjects (age: 57.6±5.1 years, 9 males) underwent 3.0 Tesla brain magnetic resonance imaging (MRI) and neuropsychological testing evaluating short- and long-term memory, executive functions, language, attention, praxia and non-verbal learning. Volumetric segmentation of cortical and subcortical structures and voxel-based morphometry (VBM) were performed. Patients and controls differed significantly in Rey Auditory-Verbal Learning test (immediate and delayed recall), Stroop test and Digit span backward scores. Volumes of cortical gray matter (GM), right hippocampus, right and left caudate were smaller in patients compared to controls, with also brain parenchymal fraction (a normalized measure of cerebral atrophy) approaching statistical significance. Differences remained significant after controlling for comorbidities (hypertension, diabetes, smoking, hypercholesterolemia). VBM analysis showed regions of decreased GM volume in right and left hippocampus and within more lateral temporal areas in patients with OSA. Our findings indicate that the significant cognitive impairment seen in patients with moderate-severe OSA is associated with brain tissue damage in regions involved in several cognitive tasks. We conclude that OSA can increase brain susceptibility to the effects of aging and other clinical and pathological occurrences.
Surgery, and specifically, tumor resection, is the primary treatment for most patients suffering from brain tumors. Medical imaging techniques, and in particular, magnetic resonance imaging are currently used in diagnosis as well as image-guided surgery procedures. However, studies show that computed tomography and magnetic resonance imaging fail to accurately identify the full extent of malignant brain tumors and their microscopic infiltration. Mass spectrometry is a well-known analytical technique used to identify molecules in a given sample based on their mass. In a recent study, it is proposed to use mass spectrometry as an intraoperative tool for discriminating tumor and non-tumor tissue. Integration of mass spectrometry with the resection module allows for tumor resection and immediate molecular analysis. In this paper, we propose a framework for tumor margin delineation using compressive sensing. Specifically, we show that the spatial distribution of tumor cell concentration can be efficiently reconstructed and updated using mass spectrometry information from the resected tissue. In addition, our proposed framework is model-free, and hence, requires no prior information of spatial distribution of the tumor cell concentration.
A longitudinal experiment was conducted to evaluate the effectiveness of new methods for learning neuroanatomy with computer-based instruction. Using a 3D graphical model of the human brain, and sections derived from the model, tools for exploring neuroanatomy were developed to encourage adaptive exploration. This is an instructional method which incorporates graphical exploration in the context of repeated testing and feedback. With this approach, 72 participants learned either sectional anatomy alone or whole anatomy followed by sectional anatomy. Sectional anatomy was explored either with perceptually continuous navigation through the sections or with discrete navigation (as in the use of an anatomical atlas). Learning was measured longitudinally to a high performance criterion. Subsequent tests examined transfer of learning to the interpretation of biomedical images and long-term retention. There were several clear results of this study. On initial exposure to neuroanatomy, whole anatomy was learned more efficiently than sectional anatomy. After whole anatomy was mastered, learners demonstrated high levels of transfer of learning to sectional anatomy and from sectional anatomy to the interpretation of complex biomedical images. Learning whole anatomy prior to learning sectional anatomy led to substantially fewer errors overall than learning sectional anatomy alone. Use of continuous or discrete navigation through sectional anatomy made little difference to measured outcomes. Efficient learning, good long-term retention, and successful transfer to the interpretation of biomedical images indicated that computer-based learning using adaptive exploration can be a valuable tool in instruction of neuroanatomy and similar disciplines.
The relationship between spatially distributed fMRI patterns and experimental stimuli or tasks offers insights into cognitive processes beyond those traceable from individual local activations. The multivariate properties of the fMRI signals allow us to infer interactions among individual regions and to detect distributed activations of multiple areas. Detection of task-specific multivariate activity in fMRI data is an important open problem that has drawn much interest recently. In this paper, we study and demonstrate the benefits of random forest classifiers and the associated Gini importance measure for selecting voxel subsets that form a multivariate neural response. The Gini importance measure quantifies the predictive power of a particular feature when considered as part of the entire pattern. The measure is based on a random sampling of fMRI time points and voxels. As a consequence the resulting voxel score, or Gini contrast, is highly reproducible and reliably includes all informative features. The method does not rely on a priori assumptions about the signal distribution, a specific statistical or functional model or regularization. Instead, it uses the predictive power of features to characterize their relevance for encoding task information. The Gini contrast offers an additional advantage of directly quantifying the task-relevant information in a multiclass setting, rather than reducing the problem to several binary classification subproblems. In a multicategory visual fMRI study, the proposed method identified informative regions not detected by the univariate criteria, such as the t-test or the F-test. Including these additional regions in the feature set improves the accuracy of multicategory classification. Moreover, we demonstrate higher classification accuracy and stability of the detected spatial patterns across runs than the traditional methods such as the recursive feature elimination used in conjunction with support vector machines.
BACKGROUND: Surgery remains the first and most important treatment modality for the majority of solid tumors. Across a range of brain tumor types and grades, postoperative residual tumor has a great impact on prognosis. The principal challenge and objective of neurosurgical intervention is therefore to maximize tumor resection while minimizing the potential for neurological deficit by preserving critical tissue.
OBJECTIVE: To introduce the integration of desorption electrospray ionization mass spectrometry into surgery for in vivo molecular tissue characterization and intraoperative definition of tumor boundaries without systemic injection of contrast agents.
METHODS: Using a frameless stereotactic sampling approach and by integrating a 3-dimensional navigation system with an ultrasonic surgical probe, we obtained image-registered surgical specimens. The samples were analyzed with ambient desorption/ionization mass spectrometry and validated against standard histopathology. This new approach will enable neurosurgeons to detect tumor infiltration of the normal brain intraoperatively with mass spectrometry and to obtain spatially resolved molecular tissue characterization without any exogenous agent and with high sensitivity and specificity.
RESULTS: Proof of concept is presented in using mass spectrometry intraoperatively for real-time measurement of molecular structure and using that tissue characterization method to detect tumor boundaries. Multiple sampling sites within the tumor mass were defined for a patient with a recurrent left frontal oligodendroglioma, World Health Organization grade II with chromosome 1p/19q codeletion, and mass spectrometry data indicated a correlation between lipid constitution and tumor cell prevalence.
CONCLUSION: The mass spectrometry measurements reflect a complex molecular structure and are integrated with frameless stereotaxy and imaging, providing 3-dimensional molecular imaging without systemic injection of any agents, which can be implemented for surgical margins delineation of any organ and with a rapidity that allows real-time analysis.
BACKGROUND: Schizophrenia is believed to result from abnormal functional integration of neural processes thought to arise from aberrant brain connectivity. However, evidence for anatomical dysconnectivity has been equivocal, and few studies have examined axonal fiber connectivity in schizophrenia at the level of whole-brain networks.
METHODS: Cortico-cortical anatomical connectivity at the scale of axonal fiber bundles was modeled as a network. Eighty-two network nodes demarcated functionally specific cortical regions. Sixty-four direction diffusion tensor-imaging coupled with whole-brain tractography was performed to map the architecture via which network nodes were interconnected in each of 74 patients with schizophrenia and 32 age- and gender-matched control subjects. Testing was performed to identify pairs of nodes between which connectivity was impaired in the patient group. The connectional architecture of patients was tested for changes in five network attributes: nodal degree, small-worldness, efficiency, path length, and clustering.
RESULTS: Impaired connectivity in the patient group was found to involve a distributed network of nodes comprising medial frontal, parietal/occipital, and the left temporal lobe. Although small-world attributes were conserved in schizophrenia, the cortex was interconnected more sparsely and up to 20% less efficiently in patients. Intellectual performance was found to be associated with brain efficiency in control subjects but not in patients.
CONCLUSIONS: This study presents evidence of widespread dysconnectivity in white-matter connectional architecture in a large sample of patients with schizophrenia. When considered from the perspective of recent evidence for impaired synaptic plasticity, this study points to a multifaceted pathophysiology in schizophrenia encompassing axonal as well as putative synaptic mechanisms.
We recently developed a functional neuroimaging technique called encephalographic magnetic resonance imaging (eMRI). Our method acquires rapid single-shot gradient-echo echo-planar MRI (repetition time=47 ms); it attempts to measure an MR signal more directly linked to neuronal electromagnetic activity than existing methods. To increase the likelihood of detecting such an MR signal, we recorded concurrent MRI and scalp electroencephalography (EEG) during fast (20-200 ms), localized, high-amplitude (>50 μV on EEG) cortical discharges in a cohort of focal epilepsy patients. Seen on EEG as interictal spikes, these discharges occur in between seizures and induced easily detectable MR magnitude and phase changes concurrent with the spikes with a lag of milliseconds to tens of milliseconds. Due to the time scale of the responses, localized changes in blood flow or hemoglobin oxygenation are unlikely to cause the MR signal changes that we observed. While the precise underlying mechanisms are unclear, in this study, we empirically investigate one potentially important confounding variable - motion. Head motion in the scanner affects both EEG and MR recording. It can produce brief "spike-like" artifacts on EEG and induce large MR signal changes similar to our interictal spike-related signal changes. In order to explore the possibility that interictal spikes were associated with head motions (although such an association had never been reported), we had previously tracked head position in epilepsy patients during interictal spikes and explicitly demonstrated a lack of associated head motion. However, that study was performed outside the MR scanner, and the root-mean-square error in the head position measurement was 0.7 mm. The large inaccuracy in this measurement therefore did not definitively rule out motion as a possible signal generator. In this study, we instructed healthy subjects to make deliberate brief (<500 ms) head motions inside the MR scanner and imaged these head motions with concurrent EEG and MRI. We compared these artifactual MR and EEG data to genuine interictal spikes. While per-voxel MR and per-electrode EEG time courses for the motion case can mimic the corresponding time courses associated with a genuine interictal spike, head motion can be unambiguously differentiated from interictal spikes via scalp EEG potential maps. Motion induces widespread changes in scalp potential, whereas interictal spikes are localized and have a regional fall-off in amplitude. These findings make bulk head motion an unlikely generator of the large spike-related MR signal changes that we had observed. Further work is required to precisely identify the underlying mechanisms.
The determination of myocardial volume at risk distal to coronary stenosis provides important information for prognosis and treatment of coronary artery disease. In this paper, we present a novel computational framework for estimating the myocardial volume at risk in computed tomography angiography (CTA) imagery. Initially, epicardial and endocardial surfaces, and coronary arteries are extracted using an active contour method. Then, the extracted coronary arteries are projected onto the epicardial surface, and each point on this surface is associated with its closest coronary artery using the geodesic distance measurement. The likely myocardial region at risk on the epicardial surface caused by a stenosis is approximated by the region in which all its inner points are associated with the sub-branches distal to the stenosis on the coronary artery tree. Finally, the likely myocardial volume at risk is approximated by the volume in between the region at risk on the epicardial surface and its projection on the endocardial surface, which is expected to yield computational savings over risk volume estimation using the entire image volume. Furthermore, we expect increased accuracy since, as compared to prior work using the Euclidean distance, we employ the geodesic distance in this work. The experimental results demonstrate the effectiveness of the proposed approach on pig heart CTA datasets.
We describe a technique that uses tractography to visualize neural pathways in human brains by extending an existing framework that uses overlapping Gaussian tensors to model the signal. At each point on the fiber, an unscented Kalman filter is used to find the most consistent direction as a mixture of previous estimates and of the local model. In our previous framework, the diffusion ellipsoid had a cylindrical shape, i.e., the diffusion tensor's second and third eigenvalues were identical. In this paper, we extend the tensor representation so that the diffusion tensor is represented by an arbitrary ellipsoid. Experiments on synthetic data show a reduction in the angular error at fiber crossings and branchings. Tests on in vivo data demonstrate the ability to trace fibers in areas containing crossings or branchings, and the tests also confirm the superiority of using a full tensor representation over the simplified model.
PURPOSE: Diffusion encoding with asymmetric gradient waveforms is appealing because the asymmetry provides superior efficiency. However, concomitant gradients may cause a residual gradient moment at the end of the waveform, which can cause significant signal error and image artifacts. The purpose of this study was to develop an asymmetric waveform designs for tensor-valued diffusion encoding that is not sensitive to concomitant gradients. METHODS: The "Maxwell index" was proposed as a scalar invariant to capture the effect of concomitant gradients. Optimization of "Maxwell-compensated" waveforms was performed in which this index was constrained. Resulting waveforms were compared to waveforms from literature, in terms of the measured and predicted impact of concomitant gradients, by numerical analysis as well as experiments in a phantom and in a healthy human brain. RESULTS: Maxwell-compensated waveforms with Maxwell indices below 100 (mT/m) ms showed negligible signal bias in both numerical analysis and experiments. By contrast, several waveforms from literature showed gross signal bias under the same conditions, leading to a signal bias that was large enough to markedly affect parameter maps. Experimental results were accurately predicted by theory. CONCLUSION: Constraining the Maxwell index in the optimization of asymmetric gradient waveforms yields efficient diffusion encoding that negates the effects of concomitant fields while enabling arbitrary shapes of the b-tensor. This waveform design is especially useful in combination with strong gradients, long encoding times, thick slices, simultaneous multi-slice acquisition, and large FOVs.
Although brain functionality is often remarkably robust to lesions and other insults, it may be fragile when these take place in specific locations. Previous attempts to quantify robustness and fragility sought to understand how the functional connectivity of brain networks is affected by structural changes, using either model-based predictions or empirical studies of the effects of lesions. We advance a geometric viewpoint relying on a notion of network curvature, the so-called Ollivier-Ricci curvature. This approach has been proposed to assess financial market robustness and to differentiate biological networks of cancer cells from healthy ones. Here, we apply curvature-based measures to brain structural networks to identify robust and fragile brain regions in healthy subjects. We show that curvature can also be used to track changes in brain connectivity related to age and autism spectrum disorder (ASD), and we obtain results that are in agreement with previous MRI studies.
The Human Placenta Project has focused attention on the need for noninvasive magnetic resonance imaging (MRI)-based techniques to diagnose and monitor placental function throughout pregnancy. The hope is that the management of placenta-related pathologies would be improved if physicians had more direct, real-time measures of placental health to guide clinical decision making. As oxygen alters signal intensity on MRI and oxygen transport is a key function of the placenta, many of the MRI methods under development are focused on quantifying oxygen transport or oxygen content of the placenta. For example, measurements from blood oxygen level-dependent imaging of the placenta during maternal hyperoxia correspond to outcomes in twin pregnancies, suggesting that some aspects of placental oxygen transport can be monitored by MRI. Additional methods are being developed to accurately quantify baseline placental oxygenation by MRI relaxometry. However, direct validation of placental MRI methods is challenging and therefore animal studies and ex vivo studies of human placentas are needed. Here we provide an overview of the current state of the art of oxygen transport and quantification with MRI. We suggest that as these techniques are being developed, increased focus be placed on ensuring they are robust and reliable across individuals and standardized to enable predictive diagnostic models to be generated from the data. The field is still several years away from establishing the clinical benefit of monitoring placental function in real time with MRI, but the promise of individual personalized diagnosis and monitoring of placental disease in real time continues to motivate this effort.
Jie Luo, Alireza Sedghi, Karteek Popuri, Dana Cobzas, Miaomiao Zhang, Frank Preiswerk, Matthew Toews, Alexandra Golby, Masashi Sugiyama, William III M Wells, and Sarah Frisken. 10/2019. “On the Applicability of Registration Uncertainty.” In MICCAI 2019, LNCS 11765: Pp. 410-9. Shenzhen, China: Springer.Abstract
Estimating the uncertainty in (probabilistic) image registration enables, e.g., surgeons to assess the operative risk based on the trustworthiness of the registered image data. If surgeons receive inaccurately calculated registration uncertainty and misplace unwarranted confidence in the alignment solutions, severe consequences may result. For probabilistic image registration (PIR), the predominant way to quantify the registration uncertainty is using summary statistics of the distribution of transformation parameters. The majority of existing research focuses on trying out different summary statistics as well as means to exploit them. Distinctively, in this paper, we study two rarely examined topics: (1) whether those summary statistics of the transformation distribution most informatively represent the registration uncertainty; (2) Does utilizing the registration uncertainty always be beneficial. We show that there are two types of uncertainties: the transformation uncertainty, Ut, and label uncertainty Ul. The conventional way of using Ut to quantify Ul is inappropriate and can be misleading. By a real data experiment, we also share a potentially critical finding that making use of the registration uncertainty may not always be an improvement.
External-beam radiotherapy followed by high dose rate (HDR) brachytherapy is the standard-of-care for treating gynecologic cancers. The enhanced soft-tissue contrast provided by magnetic resonance imaging (MRI) makes it a valuable imaging modality for diagnosing and treating these cancers. However, in contrast to computed tomography (CT) imaging, the appearance of the brachytherapy catheters, through which radiation sources are inserted to reach the cancerous tissue later on, is often variable across images. This paper reports, for the first time, a new deep-learning-based method for fully automatic segmentation of multiple closely spaced brachytherapy catheters in intraoperative MRI. Represented in the data are 50 gynecologic cancer patients treated by MRI-guided HDR brachytherapy. For each patient, a single intraoperative MRI was used. 826 catheters in the images were manually segmented by an expert radiation physicist who is also a trained radiation oncologist. The number of catheters in a patient ranged between 10 and 35. A deep 3D convolutional neural network (CNN) model was developed and trained. In order to make the learning process more robust, the network was trained 5 times, each time using a different combination of shown patients. Finally, each test case was processed by the five networks and the final segmentation was generated by voting on the obtained five candidate segmentations. 4-fold validation was executed and all the patients were segmented. An average distance error of 2.0 ± 3.4 mm was achieved. False positive and false negative catheters were 6.7% and 1.5% respectively. Average Dice score was equal to 0.60 ± 0.17. The algorithm is available for use in the open source software platform 3D Slicer allowing for wide scale testing and research discussion. In conclusion, to the best of our knowledge, fully automatic segmentation of multiple closely spaced catheters from intraoperative MR images was achieved for the first time in gynecological brachytherapy.