Skip to main content

Über dieses Buch

This book constitutes the refereed proceedings of two workshops held at the 19th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016, in Athens, Greece, in October 2016: the First Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2016, and the Second International Workshop on Deep Learning in Medical Image Analysis, DLMIA 2016. The 28 revised regular papers presented in this book were carefully reviewed and selected from a total of 52 submissions. The 7 papers selected for LABELS deal with topics from the following fields: crowd-sourcing methods; active learning; transfer learning; semi-supervised learning; and modeling of label uncertainty.The 21 papers selected for DLMIA span a wide range of topics such as image description; medical imaging-based diagnosis; medical signal-based diagnosis; medical image reconstruction and model selection using deep learning techniques; meta-heuristic techniques for fine-tuning parameter in deep learning-based architectures; and applications based on deep learning techniques.



Erratum to: Automated Retinopathy of Prematurity Case Detection with Convolutional Neural Networks

Daniel E. Worrall, Clare M. Wilson, Gabriel J. Brostow

Deep Learning in Medical Image Analysis


HEp-2 Cell Classification Using K-Support Spatial Pooling in Deep CNNs

This study addresses the recognition problem of the HEp-2 cell using indirect immunofluorescent (IIF) image analysis, which can facilitate the diagnosis of many autoimmune diseases by finding antibodies in the patient serum. Recently, a lot of automatic HEp-2 cell classification strategies including both shallow and deep methods have been developed, wherein the deep Convolutional Neural Networks (CNNs) have been proven to achieve impressive performance. However, the deep CNNs in general requires a fixed size of image as the input. In order to conquer the limitation of the fixed size problem, a spatial pyramid pooling (SPP) strategy has been proposed in general object recognition and detection. The SPP-net usually exploit max pooling strategies for aggregating all activated status of a specific neuron in a predefined spatial region by only taking the maximum activation, which achieved superior performance compared with mean pooling strategy in the traditional state-of-the-art coding methods such as sparse coding, linear locality-constrained coding and so on. However, the max pooling strategy in SPP-net only retains the strongest activated pattern, and would completely ignore the frequency: an important signature for identifying different types of images, of the activated patterns. Therefore, this study explores a generalized spatial pooling strategy, called K-support spatial pooling, in deep CNNs by integrating not only the maximum activated magnitude but also the response magnitude of the relatively activated patterns of a specific neuron together. This proposed K-support spatial pooling strategy in deep CNNs combines the popularly applied mean and max pooling methods, and then avoid awfully emphasizing of the maximum activation but preferring a group of activations in a supported region. The deep CNNs with the proposed K-support spatial pooling is applied for HEp-2 cell classification, and achieve promising performance compared with the state-of-the-art approaches.
Xian-Hua Han, Jianmei Lei, Yen-Wei Chen

Robust 3D Organ Localization with Dual Learning Architectures and Fusion

We present a robust algorithm for organ localization from 3D volumes in the presence of large anatomical and contextual variations. The 3D spatial search space is decomposed into two components: slice and pixel, both are modeled in 2D space. For each component, we adopt different learning architectures to leverage respective modeling power on global and local context at three orthogonal orientations. Unlike conventional patch-based scanning schemes in learning-based object detection algorithms, slice scanning along each orientation is applied, which significantly reduces the number of model evaluations. Object search evidence obtained from three orientations and different learning architectures is consolidated through fusion schemes to lead to the target organ location. Experiments conducted using 499 patient CT body scans show promise and robustness of the proposed approach.
Xiaoguang Lu, Daguang Xu, David Liu

Cell Segmentation Proposal Network for Microscopy Image Analysis

Accurate cell segmentation is vital for the development of reliable microscopy image analysis methods. It is a very challenging problem due to low contrast, weak boundaries, and conjoined and overlapping cells; producing many ambiguous regions, which lower the performance of automated segmentation methods. Cell proposals provide an efficient way of exploiting both spatial and temporal context, which can be very helpful in many of these ambiguous regions. However, most proposal based microscopy image analysis methods rely on fairly simple proposal generation stage, limiting their performance. In this paper, we propose a convolutional neural network based method which provides cell segmentation proposals, which can be used for cell detection, segmentation and tracking. We evaluate our method on datasets from histology, fluorescence and phase contrast microscopy and show that it outperforms state of the art cell detection and segmentation methods.
Saad Ullah Akram, Juho Kannala, Lauri Eklund, Janne Heikkilä

Vessel Detection in Ultrasound Images Using Deep Convolutional Neural Networks

Deep convolutional neural networks have achieved great results on image classification problems. In this paper, a new method using a deep convolutional neural network for detecting blood vessels in B-mode ultrasound images is presented. Automatic blood vessel detection may be useful in medical applications such as deep venous thrombosis detection, anesthesia guidance and catheter placement. The proposed method is able to determine the position and size of the vessels in images in real-time. 12,804 subimages of the femoral region from 15 subjects were manually labeled. Leave-one-subject-out cross validation was used giving an average accuracy of 94.5 %, a major improvement from previous methods which had an accuracy of 84 % on the same dataset. The method was also validated on a dataset of the carotid artery to show that the method can generalize to blood vessels on other regions of the body. The accuracy on this dataset was 96 %.
Erik Smistad, Lasse Løvstakken

Convolutional Neural Network for Reconstruction of 7T-like Images from 3T MRI Using Appearance and Anatomical Features

The advanced 7 Tesla (7T) Magnetic Resonance Imaging (MRI) scanners provide images with higher resolution anatomy than 3T MRI scanners, thus facilitating early diagnosis of brain diseases. However, 7T MRI scanners are less accessible, compared to the 3T MRI scanners. This motivates us to reconstruct 7T-like images from 3T MRI. We propose a deep architecture for Convolutional Neural Network (CNN), which uses the appearance (intensity) and anatomical (labels of brain tissues) features as input to non-linearly map 3T MRI to 7T MRI. In the training step, we train the CNN by feeding it with both appearance and anatomical features of the 3T patch. This outputs the intensity of center voxel in the corresponding 7T patch. In the testing step, we apply the trained CNN to map each input 3T patch to the 7T-like image patch. Our performance is evaluated on 15 subjects, each with both 3T and 7T MR images. Both visual and numerical results show that our method outperforms the comparison methods.
Khosro Bahrami, Feng Shi, Islem Rekik, Dinggang Shen

Fast Predictive Image Registration

We present a method to predict image deformations based on patch-wise image appearance. Specifically, we design a patch-based deep encoder-decoder network which learns the pixel/voxel-wise mapping between image appearance and registration parameters. Our approach can predict general deformation parameterizations, however, we focus on the large deformation diffeomorphic metric mapping (LDDMM) registration model. By predicting the LDDMM momentum-parameterization we retain the desirable theoretical properties of LDDMM, while reducing computation time by orders of magnitude: combined with patch pruning, we achieve a \(1500\mathsf {x}\)/\(66\mathsf {x}\) speed-up compared to GPU-based optimization for 2D/3D image registration. Our approach has better prediction accuracy than predicting deformation or velocity fields and results in diffeomorphic transformations. Additionally, we create a Bayesian probabilistic version of our network, which allows evaluation of deformation field uncertainty through Monte Carlo sampling using dropout at test time. We show that deformation uncertainty highlights areas of ambiguous deformations. We test our method on the OASIS brain image dataset in 2D and 3D.
Xiao Yang, Roland Kwitt, Marc Niethammer

Longitudinal Multiple Sclerosis Lesion Segmentation Using Multi-view Convolutional Neural Networks

Automatic segmentation of Multiple Sclerosis (MS) lesions is a challenging task due to their variability in shape, size, location and texture in Magnetic Resonance (MR) images. A reliable, automatic segmentation method can help diagnosis and patient follow-up while reducing the time consuming need of manual segmentation. In this paper, we present a fully automated method for MS lesion segmentation. The proposed method uses MR intensities and White Matter (WM) priors for extraction of candidate lesion voxels and uses Convolutional Neural Networks for false positive reduction. Our networks process longitudinal data, a novel contribution in the domain of MS lesion analysis. The method was tested on the ISBI 2015 dataset and obtained state-of-the-art Dice results with the performance level of a trained human rater.
Ariel Birenbaum, Hayit Greenspan

Automated Retinopathy of Prematurity Case Detection with Convolutional Neural Networks

Retinopathy of Prematurity (ROP) is an ocular disease observed in premature babies, considered one of the largest preventable causes of childhood blindness. Problematically, the visual indicators of ROP are not well understood and neonatal fundus images are usually of poor quality and resolution. We investigate two ways to aid clinicians in ROP detection using convolutional neural networks (CNN): (1) We fine-tune a pretrained GoogLeNet as a ROP detector and with small modifications also return an approximate Bayesian posterior over disease presence. To the best of our knowledge, this is the first completely automated ROP detection system. (2) To further aid grading, we train a second CNN to return novel feature map visualizations of pathologies, learned directly from the data. These feature maps highlight discriminative information, which we believe may be used by clinicians with our classifier to aid in screening.
Daniel E. Worrall, Clare M. Wilson, Gabriel J. Brostow

Fully Convolutional Network for Liver Segmentation and Lesions Detection

In this work we explore a fully convolutional network (FCN) for the task of liver segmentation and liver metastases detection in computed tomography (CT) examinations. FCN has proven to be a very powerful tool for semantic segmentation. We explore the FCN performance on a relatively small dataset and compare it to patch based CNN and sparsity based classification schemes. Our data contains CT examinations from 20 patients with overall 68 lesions and 43 livers marked in one slice and 20 different patients with a full 3D liver segmentation. We ran 3-fold cross-validation and results indicate superiority of the FCN over all other methods tested. Using our fully automatic algorithm we achieved true positive rate of 0.86 and 0.6 false positive per case which are very promising and clinically relevant results.
Avi Ben-Cohen, Idit Diamant, Eyal Klang, Michal Amitai, Hayit Greenspan

Deep Learning of Brain Lesion Patterns for Predicting Future Disease Activity in Patients with Early Symptoms of Multiple Sclerosis

Multiple sclerosis (MS) is a neurological disease with an early course that is characterized by attacks of clinical worsening, separated by variable periods of remission. The ability to predict the risk of attacks in a given time frame can be used to identify patients who are likely to benefit from more proactive treatment. In this paper, we aim to determine whether deep learning can extract, from segmented lesion masks, latent features that can predict short-term disease activity in patients with early MS symptoms more accurately than lesion volume, which is a very commonly used MS imaging biomarker. More specifically, we use convolutional neural networks to extract latent MS lesion patterns that are associated with early disease activity using lesion masks computed from baseline MR images. The main challenges are that lesion masks are generally sparse and the number of training samples is small relative to the dimensionality of the images. To cope with sparse voxel data, we propose utilizing the Euclidean distance transform (EDT) for increasing information density by populating each voxel with a distance value. To reduce the risk of overfitting resulting from high image dimensionality, we use a synergistic combination of downsampling, unsupervised pretraining, and regularization during training. A detailed analysis of the impact of EDT and unsupervised pretraining is presented. Using the MRIs from 140 subjects in a 7-fold cross-validation procedure, we demonstrate that our prediction model can achieve an accuracy rate of 72.9 % (SD = 10.3 %) over 2 years using baseline MR images only, which is significantly higher than the 65.0 % (SD = 14.6 %) that is attained with the traditional MRI biomarker of lesion load.
Youngjin Yoo, Lisa W. Tang, Tom Brosch, David K. B. Li, Luanne Metz, Anthony Traboulsee, Roger Tam

De-noising of Contrast-Enhanced MRI Sequences by an Ensemble of Expert Deep Neural Networks

Dynamic contrast-enhanced MRI (DCE-MRI) is an imaging protocol where MRI scans are acquired repetitively throughout the injection of a contrast agent. The analysis of dynamic scans is widely used for the detection and quantification of blood brain barrier (BBB) permeability. Extraction of the pharmacokinetic (PK) parameters from the DCE-MRI washout curves allows quantitative assessment of the BBB functionality. Nevertheless, curve fitting required for the analysis of DCE-MRI data is error-prone as the dynamic scans are subject to non-white, spatially-dependent and anisotropic noise that does not fit standard noise models. The two existing approaches i.e. curve smoothing and image de-noising can either produce smooth curves but cannot guaranty fidelity to the PK model or cannot accommodate the high variability in noise statistics in time and space.
We present a novel framework based on Deep Neural Networks (DNNs) to address the DCE-MRI de-noising challenges. The key idea is based on an ensembling of expert DNNs, where each is trained for different noise characteristics and curve prototypes to solve an inverse problem on a specific subset of the input space. The most likely reconstruction is then chosen using a classifier DNN. As ground-truth (clean) signals for training are not available, a model for generating realistic training sets with complex nonlinear dynamics is presented. The proposed approach has been applied to DCE-MRI scans of stroke and brain tumor patients and is shown to favorably compare to state-of-the-art de-noising methods, without degrading the contrast of the original images.
Ariel Benou, Ronel Veksler, Alon Friedman, Tammy Riklin Raviv

Three-Dimensional CT Image Segmentation by Combining 2D Fully Convolutional Network with 3D Majority Voting

We propose a novel approach for automatic segmentation of anatomical structures on 3D CT images by voting from a fully convolutional network (FCN), which accomplishes an end-to-end, voxel-wise multiple-class classification to map each voxel in a CT image directly to an anatomical label. The proposed method simplifies the segmentation of the anatomical structures (including multiple organs) in a CT image (generally in 3D) to majority voting for the semantic segmentation of multiple 2D slices drawn from different viewpoints with redundancy. An FCN consisting of “convolution” and “de-convolution” parts is trained and re-used for the 2D semantic image segmentation of different slices of CT scans. All of the procedures are integrated into a simple and compact all-in-one network, which can segment complicated structures on differently sized CT images that cover arbitrary CT scan regions without any adjustment. We applied the proposed method to segment a wide range of anatomical structures that consisted of 19 types of targets in the human torso, including all the major organs. A database consisting of 240 3D CT scans and a humanly annotated ground truth was used for training and testing. The results showed that the target regions for the entire set of CT test scans were segmented with acceptable accuracies (89 % of total voxels were labeled correctly) against the human annotations. The experimental results showed better efficiency, generality, and flexibility of this end-to-end learning approach on CT image segmentations comparing to conventional methods guided by human expertise.
Xiangrong Zhou, Takaaki Ito, Ryosuke Takayama, Song Wang, Takeshi Hara, Hiroshi Fujita

Medical Image Description Using Multi-task-loss CNN

Automatic detection and classification of lesions in medical images remains one of the most important and challenging problems. In this paper, we present a new multi-task convolutional neural network (CNN) approach for detection and semantic description of lesions in diagnostic images. The proposed CNN-based architecture is trained to generate and rank rectangular regions of interests (ROI’s) surrounding suspicious areas. The highest score candidates are fed into the subsequent network layers. These layers are trained to generate semantic description of the remaining ROI’s.
During the training stage, our approach uses rectangular ground truth boxes; it does not require accurately delineated lesion contours. It has a clear advantage for supervised training on large datasets. Our system learns discriminative features which are shared in the Detection and the Description stages. This eliminates the need for hand-crafted features, and allows application of the method to new modalities and organs with minimal overhead. The proposed approach generates medical report by estimating standard radiological lexicon descriptors which are a basis for diagnosis. The proposed approach should help radiologists to understand a diagnostic decision of a computer aided diagnosis (CADx) system. We test the proposed method on proprietary and publicly available breast databases, and show that our method outperforms the competing approaches.
Pavel Kisilev, Eli Sason, Ella Barkan, Sharbell Hashoul

Fully Automating Graf’s Method for DDH Diagnosis Using Deep Convolutional Neural Networks

Developmental dysplasia of the hip (DDH) is a condition affecting up to 1 in 30 infants. DDH is easy to treat if diagnosed early, but undiagnosed DDH can result in life-long hip pain, dysfunction and an increased risk of early onset osteoarthritis, and accounts for around 30 % of all hip replacements in patients under 60. The gold standard for diagnosis in infants is an ultrasound scan, followed by an analysis procedure known as Graf’s method. The application of Graf’s method is notoriously operator-dependent, requiring years of training to reach reasonable and reproducible performance. We describe a novel deep-learning based pipeline that applies Graf’s method to ultrasound scans of the hip. We use a convolutional network with an adversarial component to segment the image into relevant landmarks, and define a set of post-processing rules to translate the segmentations into Graf’s metrics. Comparing our pipeline to estimates made by experts in DDH diagnosis shows promising results.
David Golan, Yoni Donner, Chris Mansi, Jacob Jaremko, Manoj Ramachandran

Multi-dimensional Gated Recurrent Units for the Segmentation of Biomedical 3D-Data

We present a supervised deep learning method to automatically segment 3D volumes of biomedical image data. The presented method takes advantage of a neural network with the main layers consisting of multi-dimensional gated recurrent units. We apply an on-the-fly data augmentation technique which allows for accurate estimations without the need for either a huge amount of training data or advanced data pre- or postprocessing. We show that our method performs amongst the leading techniques on a popular brain segmentation challenge dataset in terms of speed, accuracy and memory efficiency. We describe in detail advantages over a similar method which uses the well-established long short-term memory.
Simon Andermatt, Simon Pezold, Philippe Cattin

Learning Thermal Process Representations for Intraoperative Analysis of Cortical Perfusion During Ischemic Strokes

Thermal imaging is a non-invasive and marker-free approach for intraoperative measurements of small temperature variations. In this work, we demonstrate the abilities of active dynamic thermal imaging for analysis of tissue perfusion state in case of cerebral ischemia. For this purpose, a NaCl irrigation is applied to the exposed cortex during hemicraniectomy. The caused temperature changes are measured by a thermal imaging system whilst tissue heating is modeled by a double exponential function. Modeled temperature decay constants allow us to characterize tissue perfusion with respect to its dynamic thermal properties. As intraoperative imaging prevents the usage of computational intense parameter optimization schemes we discuss a deep learning framework that approximates these constants given a simple temperature sequence. The framework is compared to common Levenberg-Marquardt based parameter optimization approaches. The proposed deep parameter approximation framework shows good performance compared to numerical optimization with random initialization. We further validated the approximated parameters by an intraoperative case suffering acute cerebral ischemia. The results indicate that even approximated temperature decay constants allow us to quantify cortical perfusion. Latter yield a standardized representation of cortical thermodynamic properties and might guide further research regarding specific intraoperative therapies and characterization of pathologies with atypical cortical perfusion.
Nico Hoffmann, Edmund Koch, Gerald Steiner, Uwe Petersohn, Matthias Kirsch

Automatic Slice Identification in 3D Medical Images with a ConvNet Regressor

Identification of anatomical regions of interest is a prerequisite in many medical image analysis tasks. We propose a method that automatically identifies a slice of interest (SOI) in 3D images with a convolutional neural network (ConvNet) regressor.
In 150 chest CT scans two reference slices were manually identified: one containing the aortic root and another superior to the aortic arch. In two independent experiments, the ConvNet regressor was trained with 100 CTs to determine the distance between each slice and the SOI in a CT. To identify the SOI, a first order polynomial was fitted through the obtained distances.
In 50 test scans, the mean distances between the reference and the automatically identified slices were 5.7 mm (4.0 slices) for the aortic root and 5.6 mm (3.7 slices) for the aortic arch.
The method shows similar results for both tasks and could be used for automatic slice identification.
Bob D. de Vos, Max A. Viergever, Pim A. de Jong, Ivana Išgum

Estimating CT Image from MRI Data Using 3D Fully Convolutional Networks

Computed tomography (CT) is critical for various clinical applications, e.g., radiotherapy treatment planning and also PET attenuation correction. However, CT exposes radiation during CT imaging, which may cause side effects to patients. Compared to CT, magnetic resonance imaging (MRI) is much safer and does not involve any radiation. Therefore, recently researchers are greatly motivated to estimate CT image from its corresponding MR image of the same subject for the case of radiotherapy planning. In this paper, we propose a 3D deep learning based method to address this challenging problem. Specifically, a 3D fully convolutional neural network (FCN) is adopted to learn an end-to-end nonlinear mapping from MR image to CT image. Compared to the conventional convolutional neural network (CNN), FCN generates structured output and can better preserve the neighborhood information in the predicted CT image. We have validated our method in a real pelvic CT/MRI dataset. Experimental results show that our method is accurate and robust for predicting CT image from MRI image, and also outperforms three state-of-the-art methods under comparison. In addition, the parameters, such as network depth and activation function, are extensively studied to give an insight for deep learning based regression tasks in our application.
Dong Nie, Xiaohuan Cao, Yaozong Gao, Li Wang, Dinggang Shen

The Importance of Skip Connections in Biomedical Image Segmentation

In this paper, we study the influence of both long and short skip connections on Fully Convolutional Networks (FCN) for biomedical image segmentation. In standard FCNs, only long skip connections are used to skip features from the contracting path to the expanding path in order to recover spatial information lost during downsampling. We extend FCNs by adding short skip connections, that are similar to the ones introduced in residual networks, in order to build very deep FCNs (of hundreds of layers). A review of the gradient flow confirms that for a very deep FCN it is beneficial to have both long and short skip connections. Finally, we show that a very deep FCN can achieve near-to-state-of-the-art results on the EM dataset without any further post-processing.
Michal Drozdzal, Eugene Vorontsov, Gabriel Chartrand, Samuel Kadoury, Chris Pal

Understanding the Mechanisms of Deep Transfer Learning for Medical Images

The ability to automatically learn task specific feature representations has led to a huge success of deep learning methods. When large training data is scarce, such as in medical imaging problems, transfer learning has been very effective. In this paper, we systematically investigate the process of transferring a Convolutional Neural Network, trained on ImageNet images to perform image classification, to kidney detection problem in ultrasound images. We study how the detection performance depends on the extent of transfer. We show that a transferred and tuned CNN can outperform a state-of-the-art feature engineered pipeline and a hybridization of these two techniques achieves 20 % higher performance. We also investigate how the evolution of intermediate response images from our network. Finally, we compare these responses to state-of-the-art image processing filters in order to gain greater insight into how transfer learning is able to effectively manage widely varying imaging regimes.
Hariharan Ravishankar, Prasad Sudhakar, Rahul Venkataramani, Sheshadri Thiruvenkadam, Pavan Annangi, Narayanan Babu, Vivek Vaidya

A Region Based Convolutional Network for Tumor Detection and Classification in Breast Mammography

This paper addresses the problem of detection and classification of tumors in breast mammograms. We introduce a novel system that integrates several modules including a breast segmentation module and a fibroglandular tissue segmentation module into a modified cascaded region-based convolutional network. The method is evaluated on a large multi-center clinical dataset and compared to ground truth annotated by expert radiologists. Preliminary experimental results show the high accuracy and efficiency obtained by the suggested network structure. As the volume and complexity of data in healthcare continues to accelerate generalizing such an approach may have a profound impact on patient care in many applications.
Ayelet Akselrod-Ballin, Leonid Karlinsky, Sharon Alpert, Sharbell Hasoul, Rami Ben-Ari, Ella Barkan

Large-Scale Annotation of Biomedical Data and Expert Label Synthesis


Early Experiences with Crowdsourcing Airway Annotations in Chest CT

Measuring airways in chest computed tomography (CT) images is important for characterizing diseases such as cystic fibrosis, yet very time-consuming to perform manually. Machine learning algorithms offer an alternative, but need large sets of annotated data to perform well. We investigate whether crowdsourcing can be used to gather airway annotations which can serve directly for measuring the airways, or as training data for the algorithms. We generate image slices at known locations of airways and request untrained crowd workers to outline the airway lumen and airway wall. Our results show that the workers are able to interpret the images, but that the instructions are too complex, leading to many unusable annotations. After excluding unusable annotations, quantitative results show medium to high correlations with expert measurements of the airways. Based on this positive experience, we describe a number of further research directions and provide insight into the challenges of crowdsourcing in medical images from the perspective of first-time users.
Veronika Cheplygina, Adria Perez-Rovira, Wieying Kuo, Harm A. W. M. Tiddens, Marleen de Bruijne

Hierarchical Feature Extraction for Nuclear Morphometry-Based Cancer Diagnosis

Cell and nuclear morphology, as observed from histopathology microscopy images, have long been known as important indicators of disease states. Due to the large amount of data, obtaining expert pathologists annotations at the individual cell level is impractical in many applications, however. Thus the majority of the approaches currently available for automated classification and cancer detection are based on utilizing the patient label for each segmented cell, and patient classification is performed by classifying single morphological exemplars (e.g. cells or subcellular features) in combination with a majority voting procedure. Here we propose a new hierarchical method for classifying sets of nuclei. The method can be interpreted as a type of multiple instance learning (MIL) method in that it embeds data from each patient into a hierarchical feature space. The feature space, and classification boundary, are alternatively optimized utilizing the support vector machine (SVM) cost function. We demonstrate the application of the method in the diagnosis of thyroid lesions and compare to existing MIL methods showing significant improvements in classification accuracy.
Chi Liu, Yue Huang, Ligong Han, John A. Ozolek, Gustavo K. Rohde

Using Crowdsourcing for Multi-label Biomedical Compound Figure Annotation

Information analysis or retrieval for images in the biomedical literature needs to deal with a large amount of compound figures (figures containing several subfigures), as they constitute probably more than half of all images in repositories such as PubMed Central, which was the data set used for the task. The ImageCLEFmed benchmark proposed among other tasks in 2015 and 2016 a multi-label classification task, which aims at evaluating the automatic classification of figures into 30 image types. This task was based on compound figures and thus the figures were distributed to participants as compound figures but also in a separated form. Therefore, the generation of a gold standard was required, so that algorithms of participants can be evaluated and compared. This work presents the process carried out to generate the multi-labels of \(\sim \,2650\) compound figures using a crowdsourcing approach. Automatic algorithms to separate compound figures into subfigures were used and the results were then validated or corrected via crowdsourcing. The image types (MR, CT, X–ray, ...) were also annotated by crowdsourcing including detailed quality control. Quality control is necessary to insure quality of the annotated data as much as possible. \(\sim \,625\) h were invested with a cost of \(\sim \,870\$\).
Alba Garcia Seco de Herrera, Roger Schaer, Sameer Antani, Henning Müller

Towards the Semantic Enrichment of Free-Text Annotation of Image Quality Assessment for UK Biobank Cardiac Cine MRI Scans

Image quality assessment is fundamental as it affects the level of confidence in any output obtained from image analysis. Clinical research imaging scans do not often come with an explicit evaluation of their quality, however reports are written associated to the patient/volunteer scans. This rich free-text documentation has the potential to provide automatic image quality assessment if efficiently processed and structured. This paper aims at showing how the use of Semantic Web technology for structuring free-text documentation can provide means for automatic image quality assessment. We aim to design and implement a semantic layer for a special dataset, the annotations made in the context of the UK Biobank Cardiac Cine MRI pilot study. This semantic layer will be a powerful tool to automatically infer or validate quality scores for clinical images and efficiently query image databases based on quality information extracted from the annotations. In this paper we motivate the need for this semantic layer, present an initial version of our ontology as well as preliminary results. The presented approach has the potential to be extended to broader projects and ultimately employed in the clinical setting.
Valentina Carapella, Ernesto Jiménez-Ruiz, Elena Lukaschuk, Nay Aung, Kenneth Fung, Jose Paiva, Mihir Sanghvi, Stefan Neubauer, Steffen Petersen, Ian Horrocks, Stefan Piechnik

Focused Proofreading to Reconstruct Neural Connectomes from EM Images at Scale

Identifying complex neural circuitry from electron microscopic (EM) images may help unlock the mysteries of the brain. However, identifying this circuitry requires time-consuming, manual tracing (proofreading) due to the size and intricacy of these image datasets, thus limiting analysis to small brain regions. Potential avenues to improve scalability include automatic image segmentation and crowdsourcing, but current efforts have had limited success. In this paper, we propose a new strategy, focused proofreading, that works with automatic segmentation and aims to limit proofreading to areas that are most impactful to the resulting circuit. We then introduce a novel workflow, which exploits biological information such as synapses, and apply it to a large fly optic lobe dataset. Our techniques achieve significant tracing speedups without sacrificing quality. Furthermore, our methodology makes proofreading more accessible and could enhance the effectiveness of crowdsourcing.
Stephen M. Plaza

Hands-Free Segmentation of Medical Volumes via Binary Inputs

We propose a novel hands-free method to interactively segment 3D medical volumes. In our scenario, a human user progressively segments an organ by answering a series of questions of the form “Is this voxel inside the object to segment?”. At each iteration, the chosen question is defined as the one halving a set of candidate segmentations given the answered questions. For a quick and efficient exploration, these segmentations are sampled according to the Metropolis-Hastings algorithm. Our sampling technique relies on a combination of relaxed shape prior, learnt probability map and consistency with previous answers. We demonstrate the potential of our strategy on a prostate segmentation MRI dataset. Through the study of failure cases with synthetic examples, we demonstrate the adaptation potential of our method. We also show that our method outperforms two intuitive baselines: one based on random questions, the other one being the thresholded probability map.
Florian Dubost, Loic Peter, Christian Rupprecht, Benjamin Gutierrez Becker, Nassir Navab

Playsourcing: A Novel Concept for Knowledge Creation in Biomedical Research

Being considered as a valid solution to the lack of ground truth data problem, crowdsourcing has recently gained a lot of attention within the biomedical domain. However, available concepts in life science domain require expert knowledge and thereby restrict the access to only very specific communities. In this paper, we go beyond state-of-the-art and present a novel concept for seamlessly embedding biomedical science into a common game canvas. Besides introducing the visual saliency concept, we thereby essentially eliminate the requirement for prior knowledge. We have further implemented a game to evaluate our novel concept in three different user studies.
Shadi Albarqouni, Stefan Matl, Maximilian Baust, Nassir Navab, Stefanie Demirci


Weitere Informationen

Premium Partner

BranchenIndex Online

Die B2B-Firmensuche für Industrie und Wirtschaft: Kostenfrei in Firmenprofilen nach Lieferanten, Herstellern, Dienstleistern und Händlern recherchieren.



Best Practices für die Mitarbeiter-Partizipation in der Produktentwicklung

Unternehmen haben das Innovationspotenzial der eigenen Mitarbeiter auch außerhalb der F&E-Abteilung erkannt. Viele Initiativen zur Partizipation scheitern in der Praxis jedoch häufig. Lesen Sie hier  - basierend auf einer qualitativ-explorativen Expertenstudie - mehr über die wesentlichen Problemfelder der mitarbeiterzentrierten Produktentwicklung und profitieren Sie von konkreten Handlungsempfehlungen aus der Praxis.
Jetzt gratis downloaden!