Search

Obstructive sleep apnea (OSA) is characterized by decreased breathing events that occur through the night, with severity reported as the apnea-hypopnea index (AHI), which is associated with certain craniofacial features. In this study, we used data from 1366 patients collected as part of Stanford Technology Analytics and Genomics in Sleep (STAGES) across 11 US and Canadian sleep clinics and analyzed 3D craniofacial scans with the goal of predicting AHI, as measured using gold standard nocturnal polysomnography (PSG). First, the algorithm detects pre-specified landmarks on mesh objects and aligns scans in 3D space. Subsequently, 2D images and depth maps are generated by rendering and rotating scans by 45-degree increments. Resulting images were stacked as channels and used as input to multi-view convolutional neural networks, which were trained and validated in a supervised manner to predict AHI values derived from PSGs. The proposed model achieved a mean absolute error of 11.38 events/hour, a Pearson correlation coefficient of 0.4, and accuracy for predicting OSA of 67% using 10-fold cross-validation. The model improved further by adding patient demographics and variables from questionnaires. We also show that the model performed at the level of three sleep medicine specialists, who used clinical experience to predict AHI based on 3D scan displays. Finally, we created topographic displays of the most important facial features used by the model to predict AHI, showing importance of the neck and chin area. The proposed algorithm has potential to serve as an inexpensive and efficient screening tool for individuals with suspected OSA.

PURPOSE To compare prospective motion correction (PMC) and retrospective motion correction (RMC) in Cartesian 3D-encoded MPRAGE scans and to investigate the effects of correction frequency and parallel imaging on the performance of RMC. METHODS Head motion was estimated using a markerless tracking system and sent to a modified MPRAGE sequence, which can continuously update the imaging FOV to perform PMC. The prospective correction was applied either before each echo train (before-ET) or at every sixth readout within the ET (within-ET). RMC was applied during image reconstruction by adjusting k-space trajectories according to the measured motion. The motion correction frequency was retrospectively increased with RMC or decreased with reverse RMC. Phantom and in vivo experiments were used to compare PMC and RMC, as well as to compare within-ET and before-ET correction frequency during continuous motion. The correction quality was quantitatively evaluated using the structural similarity index measure with a reference image without motion correction and without intentional motion. RESULTS PMC resulted in superior image quality compared to RMC both visually and quantitatively. Increasing the correction frequency from before-ET to within-ET reduced the motion artifacts in RMC. A hybrid PMC and RMC correction, that is, retrospectively increasing the correction frequency of before-ET PMC to within-ET, also reduced motion artifacts. Inferior performance of RMC compared to PMC was shown with GRAPPA calibration data without intentional motion and without any GRAPPA acceleration. CONCLUSION Reductions in local Nyquist violations with PMC resulted in superior image quality compared to RMC. Increasing the motion correction frequency to within-ET reduced the motion artifacts in both RMC and PMC.

We propose a pipeline for the characterization of facial and cochlear nerves in CT scans, a task specifically relevant for cochlear implant surgery planning. These structures are hard to locate in clinical CT scans due to their small size relative to the image resolution, the lack of contrast, and the proximity to other similar structures in this region. We define key landmarks around the facial and cochlear nerves and locate them using deep reinforcement learning with communicative multi-agents based on the C-MARL model. These landmarks are used as initialization for customized characterization methods. These include the automated direct measurement of the diameter of the cochlear nerve canal and extraction of the cochlear nerve cross-section followed by its segmentation using active contours. We also derive a path selection algorithm for optimal geodesic pathfinding selection based on Dijkstra’s algorithm for the characterization of the facial nerve. A total of 119 clinical CT images from preoperative patients have been used to develop this pipeline that produces accurate characterizations of these nerves in the cochlear region and provides reliable measurements for computer-aided diagnosis and surgery planning.

Patient-specific computational fluid dynamics (CFD) simulations can provide invaluable insight into the interaction of left atrial appendage (LAA) morphology, hemodynamics, and the formation of thrombi in atrial fibrillation (AF) patients. Nonetheless, CFD solvers are notoriously time-consuming and computationally demanding, which has sparked an ever-growing body of literature aiming to develop surrogate models of fluid simulations based on neural networks. The present study aims at developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), an in-silico index linked to the risk of thrombosis, typically derived from CFD simulations, solely from the patient-specific LAA morphology. To this end, a set of popular DL approaches were evaluated, including fully connected networks (FCN), convolutional neural networks (CNN), and geometric deep learning. While the latter directly operated over non-Euclidean domains, the FCN and CNN approaches required previous registration or 2D mapping of the input LAA mesh. First, the superior performance of the graph-based DL model was demonstrated in a dataset consisting of 256 synthetic and real LAA, where CFD simulations with simplified boundary conditions were run. Subsequently, the adaptability of the geometric DL model was further proven in a more realistic dataset of 114 cases, which included the complete patient-specific LA and CFD simulations with more complex boundary conditions. The resulting DL framework successfully predicted the overall distribution of the ECAP in both datasets, based solely on anatomical features, while reducing computational times by orders of magnitude compared to conventional CFD solvers. This work was supported by the Agency for Management of University and Research Grants of the Generalitat de Catalunya under the Grants for the Contracting of New Research Staff Programme—FI (2020-FI-B-00690) and the Spanish Ministry of Economy and Competitiveness under the Programme for the Formation of Doctors (PRE2018-084062), the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502) and the Retos Investigación project (RTI2018-101193-B-I00). Additionally, this work was supported by the H2020 EU SimCardioTest project (Digital transformation in Health and Care SC1-DTH-06-2020; grant agreement No. 101016496).

The task of 3D shape classification is closely related to finding a good representation of the shapes. In this study, we focus on surface representations of complex anatomies and on how such representations can be utilized for super- and unsupervised classification. We present a novel Implicit Neural Distance Representation based on unsigned distance fields (UDFs). The UDFs can be embedded into a low-dimensional latent space, which is optimized using only the shape itself. We demonstrate that this self-optimized latent space holds important global shape information useful for reconstructing the anatomies, but also that unsupervised clustering of the latent vectors successfully separates different anatomies (left atrium, left/right ear-canals and human faces). Finally, we show how the representation can be used to do gender classification of human face geometries, which is a notoriously hard problem.

Comunicació presentada a: FIMH 2021 11th International Conference, celebrada del 21 al 25 de juny de 2021 a Stanford, CA, USA. The assessment of left atrial appendage (LAA) thrombogenesis has experienced major advances with the adoption of patient-specific computational fluid dynamics (CFD) simulations. Nonetheless, due to the vast computational resources and long execution times required by fluid dynamics solvers, there is an ever-growing body of work aiming to develop surrogate models of fluid flow simulations based on neural networks. The present study builds on this foundation by developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), linked to the risk of thrombosis, solely from the patient-specific LAA geometry. To this end, we leveraged recent advancements in Geometric DL, which seamlessly extend the unparalleled potential of convolutional neural networks (CNN), to non-Euclidean data such as meshes. The model was trained with a dataset combining 202 synthetic and 54 real LAA, predicting the ECAP distributions instantaneously, with an average mean absolute error of 0.563. Moreover, the resulting framework manages to predict the anatomical features related to higher ECAP values even when trained exclusively on synthetic cases. This work was supported by the Agency for Management of University and Research Grants of the Generalitat de Catalunya under the the Grants for the Contracting of New Research Staff Programme - FI (2020 FI B 00608) and the Spanish Ministry of Economy and Competitiveness under the Programme for the Formation of Doctors (PRE2018-084062), the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502) and the Retos Investigaci´on project (RTI2018-101193-B-I00). Additionally, this work was supported by the H2020 EU SimCardioTest project (Digital transformation in Health and Care SC1- DTH-06-2020; grant agreement No. 101016496).

Background: Patient head motion is a major concern in clinical brain MRI, as it reduces the diagnostic image quality and may increase examination time and cost. Purpose: To investigate the prevalence of MR images with significant motion artifacts on a given clinical scanner and to estimate the potential financial cost savings of applying motion correction to clinical brain MRI examinations. Study Type: Retrospective. Subjects: In all, 173 patients undergoing a PET/MRI dementia protocol and 55 pediatric patients undergoing a PET/MRI brain tumor protocol. The total scan time of the two protocols were 17 and 40 minutes, respectively. Field Strength/Sequences: 3 T, Siemens mMR Biograph, MPRAGE, DWI, T 1 and T 2-weighted FLAIR, T 2-weighted 2D-FLASH, T 2-weighted TSE. Assessment: A retrospective review of image sequences from a given clinical MRI scanner was conducted to investigate the prevalence of motion-corrupted images. The review was performed by three radiologists with different levels of experience using a three-step semiquantitative scale to classify the quality of the images. A total of 1013 sequences distributed on 228 MRI examinations were reviewed. The potential cost savings of motion correction were estimated by a cost estimation for our country with assumptions. Statistical Test: The cost estimation was conducted with a 20% lower and upper bound on the model assumptions to include the uncertainty of the assumptions. Results: 7.9% of the sequences had motion artifacts that decreased the interpretability, while 2.0% of the sequences had motion artifacts causing the images to be nondiagnostic. The estimated annual cost to the clinic/hospital due to patient head motion per scanner was $45,066 without pediatric examinations and $364,242 with pediatric examinations. Data Conclusion: The prevalence of a motion-corrupted image was found in 2.0% of the reviewed sequences. Based on the model, repayment periods are presented as a function of the price for applying motion correction and its performance. Evidence Level: 4. Technical Efficacy: Stage 6 J. Magn. Reson. Imaging 2020;52:731–738.

Severe malaria is mostly caused by Plasmodium falciparum, resulting in considerable, systemic inflammation and pronounced endothelial activation. The endothelium forms an interface between blood and tissue, and vasculopathy has previously been linked with malaria severity. We studied the extent to which the endothelial glycocalyx that normally maintains endothelial function is involved in falciparum malaria pathogenesis by using incident dark-field imaging in the buccal mucosa. This enabled calculation of the perfused boundary region, which indicates to what extent erythrocytes can permeate the endothelial glycocalyx. The perfused boundary region was significantly increased in severe malaria patients and mirrored by an increase of soluble glycocalyx components in plasma. This is suggestive of a substantial endothelial glycocalyx loss. Patients with severe malaria had significantly higher plasma levels of sulfated glycosaminoglycans than patients with uncomplicated malaria, whereas other measured glycocalyx markers were raised to a comparable extent in both groups. In severe malaria, the plasma level of the glycosaminoglycan hyaluronic acid was positively correlated with the perfused boundary region in the buccal cavity. Plasma hyaluronic acid and heparan sulfate were particularly high in severe malaria patients with a low Blantyre coma score, suggesting involvement in its pathogenesis. In vivo imaging also detected perivascular hemorrhages and sequestering late-stage parasites. In line with this, plasma angiopoietin-1 was decreased while angiopoietin-2 was increased, suggesting vascular instability. The density of hemorrhages correlated negatively with plasma levels of angiopoietin-1. Our findings indicate that as with experimental malaria, the loss of endothelial glycocalyx is associated with vascular dysfunction in human malaria and is related to severity.

Images are normally stored on storage devices like the internal memory of a camera or on a hard disk. To be able to store an image, an image format needs to be defined and a method to compress the image. In this chapter, a brief overview of image storage formats and compression is given.

Whole heart segmentation from cardiac CT scans is a prerequisite for many clinical applications, but manualdelineation is a tedious task and subject to both intra- and inter-observer variation. Automating the segmentation process has thus become an increasingly popular task in the field of image analysis, and is generally solved by either using 3D methods, considering the image volume as a whole, or 2D methods, segmenting each slice independently. In the field of deep learning, there are significant limitations regarding 3D networks, including the need for more training examples and GPU memory. The need for GPU memory is usually solved by down sampling the input images, thus losing important information, which is not a necessary sacrifice when employing 2D networks. It would therefore be relevant to exploit the benefits of 2D networks in a configuration, where spatial information across slices is kept, as when employing 3D networks. The proposed method performs multiclass segmentation of cardiac CT scans utilizing 2D convolutional neural networks with a multi-planar approach. Furthermore, spatial propagation is included in the network structure, to ensure spatial consistency through each image volume. The approach keeps the computational assets of 2D methods while addressing 3D issues regarding spatial context. The pipeline is structured in a two-step approach, in which the first step detects the location of the heart and crops a region of interest, and the second step performs multi-class segmentation of the heart structures. The pipeline demonstrated promising results on the MICCAI 2017 Multi-Modality Whole Heart Segmentation challenge data.

An image analysis problem can often be described as a task of recognizing patterns in images. Example patterns are straight lines and curve-like structures. The Hough transform is a highly effective algorithm for detecting straight lines in images and is the topic of the first part of the chapter. In the second part, a method for tracing paths in images is presented. The algorithm is very fast and is based on the algorithmic approach called dynamic programming.

Recently, the risk of thrombus formation in the left atrium (LA) has been assessed through patient-specific computational fluid dynamic (CFD) simulations, characterizing the complex 4D nature of blood flow in the left atrial appendage (LAA). Nevertheless, the vast computational resources and long computing times required by traditional CFD methods prevents its embedding in the clinical workflow of time-sensitive applications. In this study, two distinct deep learning (DL) architectures have been developed to receive the patient-specific LAA geometry as an input and predict the endothelial cell activation potential (ECAP), which is linked to the risk of thrombosis. The first network is based on a simple fully-connected network, while the latter also performs a dimensionality reduction of the variables. Both models have been trained with a synthetic dataset of 210 LAA geometries being able to accurately predict the ECAP distributions with an average error of 4.72% for the fully-connected approach and 5.75% for its counterpart. Most importantly, the obtention of the ECAP predictions was quasi-instantaneous, orders of magnitude faster than conventional CFD.

BACKGROUND The assessment of motor disturbances in antipsychotic-treated adolescent patients is often limited to the use of observer-based rating scales with interobserver variability. The objectives of this pilot study were to measure movement patterns associated with antipsychotic-induced parkinsonism in young patients with psychosis and initiating/treated with antipsychotics, using a computer application connected with the Microsoft Kinect sensor (Motorgame). METHOD All participants were assessed by neurological examination, clinical side effect rating scales (Udvalg for Kliniske Undersogelser Side Effect Rating Scale, Barnes Akathisia Rating Scale, Simpson Angus Scale (SAS), and Abnormal Involuntary Movement Scale), and the Motorgame. Furthermore, speed of information processing and motor speed with subtests from the Brief Assessment of Cognition in Schizophrenia test battery was assessed. RESULTS We included 21 adolescents with first-episode psychosis (62% treated with antipsychotics; males 38%; mean age 16 ± 1.4 years) and 69 healthy controls (males 36%; mean age 16 ± 1.5 years). Prolonged time of motor performance (TOMP) in the Motorgame was associated with higher SAS scores for arm dropping (p = .009). A consistent practice effect was detected (p

Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.

Timely detection of surface damages on wind turbine blades is imperative for minimising downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analysed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost, thereby reducing the overall maintenance cost arising from the manual labour involved. In this work, we develop a deep learning based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach could achieve almost human level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets advanced data augmentation during deep learning training can better generalise the trained model providing a significant gain in precision.

Morphological analysis of the left atrial appendage is an important tool to assess risk of ischemic stroke. Most deep learning approaches for 3D segmentation is guided by binary labelmaps, which results in voxelized segmentations unsuitable for morphological analysis. We propose to use signed distance fields to guide a deep network towards morphologically consistent 3D models. The proposed strategy is evaluated on a synthetic dataset of simple geometries, as well as a set of cardiac computed tomography images containing the left atrial appendage. The proposed method produces smooth surfaces with a closer resemblance to the true surface in terms of segmentation overlap and surface distance., Comment: MIDL 2019 [arXiv:1907.08612]

Fifteen percent of all strokes are caused by emboli formed in the left atrium (LA) in case of atrial fibrillation (AF). The most common site of thrombus formation is inside the left atrial appendage (LAA). The LAA is accounting for 70% to 90% of the thrombi formed in the LA in patients with non-valvular AF. Studies have shown there is a correlation between the LAA morphology and risk of ischemic stroke; Chicken Wing and Cauliflower LAA shapes are associated with lower and higher risk, respectively. These two LAA shape categories come from a popular classification in the medical domain, but it is subjective and based on qualitative shape parameters. In this paper, we describe a full framework for shape analysis and clustering of the LAA. Initially, we build a point distribution model to quantitatively describe the LAA shape variation based on 103 LAA surfaces segmented and reconstructed from multidetector computed tomography volumes. We are successfully able to determine point correspondence between LAA surfaces, by non-rigid volumetric registration of signed distance fields. To validate if LAA shapes are clustered, we employ an unsupervised clustering on the shape models parameters to estimate the natural number of clusters in our training set, where the number of shape clusters is estimated by validating the test log-likelihood of several Gaussian mixture models using two level cross-validation. We found that the LAAs surfaces basically formed two shape clusters broadly corresponding to the Chicken wing and non-Chicken Wing morphologies, which fits well with clinical knowledge.

The rapid increase in the availability of accurate 3D scanning devices has moved facial recognition and analysis into the 3D domain. 3D facial landmarks are often used as a simple measure of anatomy and it is crucial to have accurate algorithms for automatic landmark placement. The current state-of-the-art approaches have yet to gain from the dramatic increase in performance reported in human pose tracking and 2D facial landmark placement due to the use of deep convolutional neural networks (CNN). Development of deep learning approaches for 3D meshes has given rise to the new subfield called geometric deep learning, where one topic is the adaptation of meshes for the use of deep CNNs. In this work, we demonstrate how methods derived from geometric deep learning, namely multi-view CNNs, can be combined with recent advances in human pose tracking. The method finds 2D landmark estimates and propagates this information to 3D space, where a consensus method determines the accurate 3D face landmark position. We utilise the method on a standard 3D face dataset and show that it outperforms current methods by a large margin. Further, we demonstrate how models trained on 3D range scans can be used to accurately place anatomical landmarks in magnetic resonance images.

International audience; Recent developments in computational modeling of cochlear implantation are promising to study in silico the performance of the implant before surgery. However, creating a complete computational model of the patient’s anatomy while including an external device geometry remains challenging. To address such a challenge, we propose an automatic framework for the generation of patient-specific meshes for finite element modeling of the implanted cochlea. First, a statistical shape model is constructed from high-resolution anatomical μCT images. Then, by fitting the statistical model to a patient’s CT image, an accurate model of the patient-specific cochlea anatomy is obtained. An algorithm based on the parallel transport frame is employed to perform the virtual insertion of the cochlear implant. Our automatic framework also incorporates the surrounding bone and nerve fibers and assigns constitutive parameters to all components of the finite element model. This model can then be used to study in silico the effects of the electrical stimulation of the cochlear implant. Results are shown on a total of 25 models of patients. In all cases, a final mesh suitable for finite element simulations was obtained, in an average time of 94 s. The framework has proven to be fast and robust, and is promising for a detailed prognosis of the cochlear implantation surgery.

Research into the importance of the human genome in the context of facial appearance is receiving increasing attention and has led to the detection of several Single Nucleotide Polymorphisms (SNPs) of importance. In this work we attempt a holistic approach predicting facial characteristics from genetic principal components across a population of 1,266 individuals. For this we perform a genome-wide association analysis to select a large number of SNPs linked to specific facial traits, recode these to genetic principal components and then use these principal components as predictors for facial traits in a linear regression. We show in this proof-of-concept study for facial trait prediction from genome-wide SNP data that some facial characteristics can be modeled by genetic information: facial width, eyebrow width, distance between eyes, and features involving mouth shape are predicted with statistical significance (p < 0.03).

ObjectiveWe demonstrate and evaluate the first markerless motion tracker compatible with PET, MRI, and simultaneous PET/MRI systems for motion correction (MC) of brain imaging.MethodsPET and MRI compatibility is achieved by careful positioning of in-bore vision extenders and by placing all electronic components out-of-bore. The motion tracker is demonstrated in a clinical setup during a pediatric PET/MRI study including 94 pediatric patient scans. PET MC is presented for two of these scans using a customized version of the Multiple Acquisition Frame method. Prospective MC of MRI acquisition of two healthy subjects is demonstrated using a motion-aware MRI sequence. Real-time motion estimates are accompanied with a tracking validity parameter to improve tracking reliability.ResultsFor both modalities, MC shows that motion induced artifacts are noticeably reduced and that motion estimates are sufficiently accurate to capture motion ranging from small respiratory motion to large intentional motion. In the PET/MRI study, a time-activity curve analysis shows image improvements for a patient performing head movements corresponding to a tumor motion of ±5-10 mm with a 19% maximal difference in standardized uptake value before and after MC.ConclusionThe first markerless motion tracker is successfully demonstrated for prospective MC in MRI and MC in PET with good tracking validity.SignificanceAs simultaneous PET/MRI systems have become available for clinical use, an increasing demand for accurate motion tracking and MC in PET/MRI scans has emerged. The presented markerless motion tracker facilitate this demand.

3D image stacks are routinely acquired to capture data that lie on undulating 3D manifolds yet processed in 2D by biologists. Algorithms to reconstruct the specimen morphology into a 2D representation from the 3D image volume are employed in such scenarios. In this paper, we present FastSME, which offers several improvements on the baseline SME algorithm which enables accurate 2D representation of data on a manifold from 3D volumes, however is computationally expensive. The improvements are achieved in terms of processing speed (3X-10X speed-up depending on image size), minimizing sensitivity to initialization, and also increases local smoothness of the recovered manifold resulting in better reconstructed 2D composite image. We compare the proposed FastSME against the baseline SME as well as other accessible state-of-the-art tools on synthetic and real microscopy data. Our evaluation on multiple metrics demonstrates the efficiency of the presented method in maintaining fidelity of manifold shape and hence specimen morphology.

Numerical simulations offer a feasible alternative to the direct acoustic measurement of individual head-related transfer functions (HRTFs). For the acquisition of high quality 3D surface scans, as required for these simulations, several approaches exist. In this paper, we systematically analyze the variations between different approaches and evaluate the influence of the accuracy of 3D scans on the resulting simulated HRTFs. To assess this effect, HRTFs were numerically simulated based on 3D scans of the head and pinna of the FABIAN dummy head generated with 6 different methods. These HRTFs were analyzed in terms of interaural time difference, interaural level difference, energetic error in auditory filters and by their modeled localization performance. From the results, it is found that a geometric precision of about 1 mm is needed to maintain accurate localization cues, while a precision of about 4 mm is sufficient to maintain the overall spectral shape.

We present a framework for assessing which types of simple movement tasks are most discriminative between healthy controls and Parkinson’s patients. We collected movement data in a game-like environment, where we used the Microsoft Kinect sensor for tracking the user’s joints. We recruited 63 individuals for the study, of whom 30 had been diagnosed with Parkinson’s disease. A physician evaluated all participants on movement-related rating scales, e.g., elbow rigidity. The participants also completed the game task, moving their arms through a specific pattern. We present an innovative approach for data acquisition in a game-like environment, and we propose a novel method, sparse ordinal regression, for predicting the severity of motion disorders from the data.

Fully automatic segmentation of intracranial volume.Equidistant subsampling with Delaunay tetrahedralisation for dense neighborhood connectivity without blocky grid-bias.Node clamping forces the segmentation to be consistent and of genus 0.Dice scores above 98% compared to manual reference. The intracranial volume (ICV) in children with premature fusion of one or more sutures in the calvaria is of interest due to the risk of increased intracranial pressure. Challenges for automatic estimation of ICV include holes in the skull e.g. the foramen magnum and fontanelles. In this paper, we present a fully automatic 3D graph-based method for segmentation of the ICV in non-contrast CT scans. We reformulate the ICV segmentation problem as an optimal genus 0 segmentation problem in a volumetric graph. The graph is the result of a volumetric spherical subsampling. The equidistantly sampled data points are connected using Delaunay tetrahedralisation creating a highly connected neighborhood. A Markov Random Field (MRF) is constructed on the graph with probabilities learned from an Expectation Maximisation algorithm matching a Mixture of Gaussians to the data. The result of the MRF segmentation is compared to manual segmentations performed by an expert. We have achieved very high Dice scores ranging from 98.14% to 99.00%, while volume deviation from the manual segmentation ranges from 0.7% to 3.7%. The Hausdorff distance, which shows the maximum error from automatic to manual segmentation, ranges from 4.73 to 9.81mm. Since this is sensitive to single error, we have also found the 95% Hausdorff distance, which ranges from 1.10 to 3.65mm. The segmentation is very consistent with the reference and differs only in difficult areas, where it seems that our method is much more slice-wise consistent than a manual segmentation. The proposed method is expected to perform well for other volumetric segmentations.

Understanding the human inner ear anatomy and its internal structures is paramount to advance hearing implant technology. While the emergence of imaging devices allowed researchers to improve understanding of intracochlear structures, the difficulties to collect appropriate data has resulted in studies conducted with few samples. To assist the cochlear research community, a large collection of human temporal bone images is being made available. This data descriptor, therefore, describes a rich set of image volumes acquired using cone beam computed tomography and micro-CT modalities, accompanied by manual delineations of the cochlea and sub-compartments, a statistical shape model encoding its anatomical variability, and data for electrode insertion and electrical simulations. This data makes an important asset for future studies in need of high-resolution data and related statistical data objects of the cochlea used to leverage scientific hypotheses. It is of relevance to anatomists, audiologists, computer scientists in the different domains of image analysis, computer simulations, imaging formation, and for biomedical engineers designing new strategies for cochlear implantations, electrode design, and others. This work was financially supported by the European Commission FP7 (HEAR-EU European project #304857 http://www.hear-eu.eu) and the Swiss National Science Foundation (Nano- Tera initiative project title Hear-Restore). In addition, this work is partly supported by the Spanish Ministry of Economy and Competitiveness under the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502).

While the acquisition and analysis of 3D faces has been an active area of research for decades, it is still a complex and demanding task to accurately model the entire head and ears. Having accurate models would for example enable virtual design of hearing devices. In this paper, we describe a complete framework for surface registration of complete human heads where the result is point correspondence with a very high number of points. The method is based on a volumetric and multi-scale non-rigid registration of signed distance fields. The method is evaluated on a set of 30 human heads and the results are convincing. The output can for example be used to compute statistical shape models. The accuracy of predicted anatomical landmarks is on the level of experienced human operators.

Medial structures (skeletons and medial manifolds) have shown capacity to describe shape in a compact way. In the field of medical imaging, they have been employed to enrich the description of organ anatomy, to improve segmentation, or to describe the organ position in relation to surrounding structures. Methods for generation of medial structures, however, are prone to the generation of medial artifacts (spurious branches) that traditionally need to be pruned before the medial structure can be used for further computations. The act of pruning can affect main sections of the medial surface, hindering its performance as shape descriptor. In this work, we present a method for the computation of medial structures that generates smooth medial surfaces that do not need to be explicitly pruned. Additionally, we present a validation framework for medial surface evaluation. Finally, we apply this method to create a parametric model of the cochlea shape that yields better registration results between cochleae.

Non-invasive food inspection and quality assurance are becoming viable techniques in food production due to the introduction of fast and accessible multispectral X-ray scanners. However, the novel devices produce massive amount of data and there is a need for fast and accurate algorithms for processing it. We apply a sparse classifier for foreign object detection and segmentation in multispectral X-ray. Using sparse methods makes it possible to potentially use fewer variables than traditional methods and thereby reduce acquisition time, data volume and classification speed. We report our results on two datasets with foreign objects, one set with spring rolls and one with minced meat. Our results indicate that it is possible to limit the amount of data stored to 50% of the original size without affecting classification accuracy of materials used for training. The method has attractive computational properties, which allows for fast classification of items in new images.

Individual head-related transfer functions (HRTFs) are essential in applications like fitting hearing-assistive devices (HADs) for providing accurate sound localization performance. Individual HRTFs are usually obtained through intricate acoustic measurements. This paper investigates the use of a three-dimensional (3D) head model for acquisition of individual HRTFs. Two aspects were investigated; whether a 3D-printed model can replace measurements on a human listener and whether numerical simulations can replace acoustic measurements. For this purpose, HRTFs were acoustically measured for four human listeners and for a 3D printed head model of one of these listeners. Further, HRTFs were simulated by applying the finite element method to the 3D head model. The monaural spectral features and spectral distortions were very similar between re-measurements and between human and printed measurements, however larger deviations were observed between measurement and simulation. The binaural cues were in agreement among all HRTFs of the same listener, indicating that the 3D model is able to provide localization cues potentially accessible to HAD users. Hence, the pipeline of geometry acquisition, printing, and acoustic measurements or simulations, seems to be a promising step forward towards in-silico design of HADs.

When implanting cochlear implants the positions of electrodes have a large impact on the quality of the restored hearing. Due to metal artifacts it is difficult to estimate the precise location in post-operative scans. In this paper we present a method for automatically locating and determining the ordering of electrode contacts on implanted electrode arrays from post-operative CT images. Our method applies a specialized filter chain to the images based on a threshold and spherical measure, and selects contact positions at local maxima in the filtered image. Two datasets of 13 temporal bone specimens scanned in CBCT are used to validate the method, which successfully locates the electrode array in every image.

Cochlear implantation is carried out to recover the sense of hearing. However, its functional outcome varies highly between patients. In the current work, we present a study to assess the functional outcomes of cochlear implants considering the inter-variability found among a population of patients. In order to capture the cochlear anatomical details, a statistical shape model is created from high-resolution human \(\mu \)CT data. A population of virtual patients is automatically generated by sampling new anatomical instances from the statistical shape model. For each virtual patient, an implant insertion is simulated and a finite element model is generated to estimate the electrical field created into the cochlea. These simulations are defined according to the monopolar stimulation protocol of a cochlear implant and a prediction of the voltage spread over the population of virtual patients is evaluated.

Medical-image analysis requires an understanding of sophisticated scanning modalities, constructing geometric models, building meshes to represent domains, and downstream biological applications. These four steps form an image-to-mesh pipeline. For research in this field to progress, the imaging, modeling, and simulation communities will need to work together more closely.

We present a system for head motion tracking in 3D brain imaging. The system is based on facial surface reconstruction and tracking using a structured light (SL) scanning principle. The system is designed to fit into narrow 3D medical scanner geometries limiting the field of view. It is tested in a clinical setting on the high resolution research tomograph (HRRT), Siemens PET scanner with a head phantom and volunteers. The SL system is compared to a commercial optical tracking system, the Polaris Vicra system, from NDI based on translatory and rotary ground truth motions of the head phantom. The accuracy of the systems was similar, with root mean square (rms) errors of 0.09 degrees for ±20 degrees axial rotations, and rms errors of 0.24 mm for ± 25 mm translations. Tests were made using (1) a light emitting diode (LED) based miniaturized video projector, the Pico projector from Texas Instruments, and (2) a customized version of this projector replacing a visible light LED with a 850 nm near infrared LED. The latter system does not provide additional discomfort by visible light projection into the patient's eyes. The main advantage over existing head motion tracking devices, including the Polaris Vicra system, is that it is not necessary to place markers on the patient. This provides a simpler workflow and eliminates uncertainties related to marker attachment and stability. We show proof of concept of a marker less tracking system especially designed for clinical use with promising results.