Leaping into the rapidly developing world of deep learning is an exciting and sometimes confusing adventure. All of the advice and tutorials available can be hard to organize and work through, especially when training specific models on specific datasets, different from those originally used to train the network. In this short guide, we aim to walk the reader through the techniques that we have used to successfully train two deep neural networks for pixel-wise classification, including some data management and augmentation approaches for working with image data that may be insufficiently annotated or relatively homogenous.
The tracing of neural pathways through large volumes of image data is an incredibly tedious and time-consuming process that significantly encumbers progress in neuroscience. We are exploring deep learning's potential to automate segmentation of high-resolution scanning electron microscope (SEM) image data to remove that barrier. We have started with neural pathway tracing through 5.1GB of whole-brain serial-section slices from larval zebrafish collected by the Center for Brain Science at Harvard University. This kind of manual image segmentation requires years of careful work to properly trace the neural pathways in an organism as small as a zebrafish larva (approximately 5mm in total body length). In automating this process, we would vastly improve productivity, leading to faster data analysis and breakthroughs in understanding the complexity of the brain. We will build upon prior attempts to employ deep learning for automatic image segmentation extending methods for unconventional deep learning data.
With the introduction of fully convolutional neural networks, deep learning has raised the benchmark for medical image segmentation on both speed and accuracy, and different networks have been proposed for 2D and 3D segmentation with promising results. Nevertheless, most networks only handle relatively small numbers of labels (<10), and there are very limited works on handling highly unbalanced object sizes especially in 3D segmentation. In this paper, we propose a network architecture and the corresponding loss function which improve segmentation of very small structures. By combining skip connections and deep supervision with respect to the computational feasibility of 3D segmentation, we propose a fast converging and computationally efficient network architecture for accurate segmentation. Furthermore, inspired by the concept of focal loss, we propose an exponential logarithmic loss which balances the labels not only by their relative sizes but also by their segmentation difficulties. We achieve an average Dice coefficient of 82% on brain segmentation with 20 labels, with the ratio of the smallest to largest object sizes as 0.14%. Less than 100 epochs are required to reach such accuracy, and segmenting a 128x128x128 volume only takes around 0.4 s.
In this paper, we propose an efficient pseudo-marginal Markov chain Monte Carlo (MCMC) sampling approach to draw samples from posterior shape distributions for image segmentation. The computation time of the proposed approach is independent from the size of the training set used to learn the shape prior distribution nonparametrically. Therefore, it scales well for very large data sets. Our approach is able to characterize the posterior probability density in the space of shapes through its samples, and to return multiple solutions, potentially from different modes of a multimodal probability density, which would be encountered, e.g., in segmenting objects from multiple shape classes. Experimental results demonstrate the potential of the proposed approach.
Learning long-term spatial-temporal features are critical for many video analysis tasks. However, existing video segmentation methods predominantly rely on static image segmentation techniques, and methods capturing temporal dependency for segmentation have to depend on pretrained optical flow models, leading to suboptimal solutions for the problem. End-to-end sequential learning to explore spatial-temporal features for video segmentation is largely limited by the scale of available video segmentation datasets, i.e., even the largest video segmentation dataset only contains 90 short video clips. To solve this problem, we build a new large-scale video object segmentation dataset called YouTube Video Object Segmentation dataset (YouTube-VOS). Our dataset contains 3,252 YouTube video clips and 78 categories including common objects and human activities. This is by far the largest video object segmentation dataset to our knowledge and we have released it at https://youtube-vos.org. Based on this dataset, we propose a novel sequence-to-sequence network to fully exploit long-term spatial-temporal information in videos for segmentation. We demonstrate that our method is able to achieve the best results on our YouTube-VOS test set and comparable results on DAVIS 2016 compared to the current state-of-the-art methods. Experiments show that the large scale dataset is indeed a key factor to the success of our model.
Contemporary deep learning based medical image segmentation algorithms require hours of annotation labor by domain experts. These data hungry deep models perform sub-optimally in the presence of limited amount of labeled data. In this paper, we present a data efficient learning framework using the recent concept of Generative Adversarial Networks; this allows a deep neural network to perform significantly better than its fully supervised counterpart in low annotation regime. The proposed method is an extension of our previous work with the addition of a new unsupervised adversarial loss and a structured prediction based architecture. To the best of our knowledge, this work is the first demonstration of an adversarial framework based structured prediction model for medical image segmentation. Though generic, we apply our method for segmentation of blood vessels in retinal fundus images. We experiment with extreme low annotation budget (0.8 - 1.6% of contemporary annotation size). On DRIVE and STARE datasets, the proposed method outperforms our previous method and other fully supervised benchmark models by significant margins especially with very low number of annotated examples. In addition, our systematic ablation studies suggest some key recipes for successfully training GAN based semi-supervised algorithms with an encoder-decoder style network architecture.
This chapter provides insight on how iris recognition, one of the leading biometric identification technologies in the world, can be impacted by pathologies and illnesses present in the eye, what are the possible repercussions of this influence, and what are the possible means for taking such effects into account when matching iris samples. To make this study possible, a special database of iris images has been used, representing more than 20 different medical conditions of the ocular region (including cataract, glaucoma, rubeosis iridis, synechiae, iris defects, corneal pathologies and other) and containing almost 3000 samples collected from 230 distinct irises. Then, with the use of four different iris recognition methods, a series of experiments has been conducted, concluding in several important observations. One of the most popular ocular disorders worldwide - the cataract - is shown to worsen genuine comparison scores when results obtained from cataract-affected eyes are compared to those coming from healthy irises. An analysis devoted to different types of impact on eye structures caused by diseases is also carried out with significant results. The enrollment process is highly sensitive to those eye conditions that make the iris obstructed or introduce geometrical distortions. Disorders affecting iris geometry, or producing obstructions are exceptionally capable of degrading the genuine comparison scores, so that the performance of the entire biometric system can be influenced. Experiments also reveal that imperfect execution of the image segmentation stage is the most prominent contributor to recognition errors.
Deep learning approaches have achieved state-of-the-art performance in cardiac magnetic resonance (CMR) image segmentation. However, most approaches have focused on learning image intensity features for segmentation, whereas the incorporation of anatomical shape priors has received less attention. In this paper, we combine a multi-task deep learning approach with atlas propagation to develop a shape-constrained bi-ventricular segmentation pipeline for short-axis CMR volumetric images. The pipeline first employs a fully convolutional network (FCN) that learns segmentation and landmark localisation tasks simultaneously. The architecture of the proposed FCN uses a 2.5D representation, thus combining the computational advantage of 2D FCNs networks and the capability of addressing 3D spatial consistency without compromising segmentation accuracy. Moreover, the refinement step is designed to explicitly enforce a shape constraint and improve segmentation quality. This step is effective for overcoming image artefacts (e.g. due to different breath-hold positions and large slice thickness), which preclude the creation of anatomically meaningful 3D cardiac shapes. The proposed pipeline is fully automated, due to network's ability to infer landmarks, which are then used downstream in the pipeline to initialise atlas propagation. We validate the pipeline on 1831 healthy subjects and 649 subjects with pulmonary hypertension. Extensive numerical experiments on the two datasets demonstrate that our proposed method is robust and capable of producing accurate, high-resolution and anatomically smooth bi-ventricular 3D models, despite the artefacts in input CMR volumes.
Every year millions of people die due to disease of Cancer. Due to its invasive nature it is very complex to cure even in primary stages. Hence, only method to survive this disease completely is via forecasting by analyzing the early mutation in cells of the patient biopsy. Cell Segmentation can be used to find cell which have left their nuclei. This enables faster cure and high rate of survival. Cell counting is a hard, yet tedious task that would greatly benefit from automation. To accomplish this task, segmentation of cells need to be accurate. In this paper, we have improved the learning of training data by our network. It can annotate precise masks on test data. we examine the strength of activation functions in medical image segmentation task by improving learning rates by our proposed Carving Technique. Identifying the cells nuclei is the starting point for most analyses, identifying nuclei allows researchers to identify each individual cell in a sample, and by measuring how cells react to various treatments, the researcher can understand the underlying biological processes at work. Experimental results shows the efficiency of the proposed work.
Existing works on semantic segmentation typically consider a small number of labels, ranging from tens to a few hundreds. With a large number of labels, training and evaluation of such task become extremely challenging due to correlation between labels and lack of datasets with complete annotations. We formulate semantic segmentation as a problem of image segmentation given a semantic concept, and propose a novel system which can potentially handle an unlimited number of concepts, including objects, parts, stuff, and attributes. We achieve this using a weakly and semi-supervised framework leveraging multiple datasets with different levels of supervision. We first train a deep neural network on a 6M stock image dataset with only image-level labels to learn visual-semantic embedding on 18K concepts. Then, we refine and extend the embedding network to predict an attention map, using a curated dataset with bounding box annotations on 750 concepts. Finally, we train an attention-driven class agnostic segmentation network using an 80-category fully annotated dataset. We perform extensive experiments to validate that the proposed system performs competitively to the state of the art on fully supervised concepts, and is capable of producing accurate segmentations for weakly learned and unseen concepts.
Within this thesis we propose a platform for combining Augmented Reality (AR) hardware with machine learning in a user-oriented pipeline, offering to the medical staff an intuitive 3D visualization of volumetric Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) medical image segmentations inside the AR headset, that does not need human intervention for loading, processing and segmentation of medical images. The AR visualization, based on Microsoft HoloLens, employs a modular and thus scalable frontend-backend architecture for real-time visualizations on multiple AR headsets. As Convolutional Neural Networks (CNNs) have lastly demonstrated superior performance for the machine learning task of image semantic segmentation, the pipeline also includes a fully automated CNN algorithm for the segmentation of the liver from CT scans. The model is based on the Deep Retinal Image Understanding (DRIU) model which is a Fully Convolutional Network with side outputs from feature maps with different resolution, extracted at different stages of the network. The algorithm is 2.5D which means that the input is a set of consecutive scan slices. The experiments have been performed on the Liver Tumor Segmentation Challenge (LiTS) dataset for liver segmentation and demonstrated good results and flexibility. While multiple approaches exist in the domain, only few of them have focused on overcoming the practical aspects which still largely hold this technology away from the operating rooms. In line with this, we also are next planning an evaluation from medical doctors and radiologists in a real-world environment.
We propose an end-to-end learning framework for segmenting generic objects in both images and videos. Given a novel image or video, our approach produces a pixel-level mask for all "object-like" regions---even for object categories never seen during training. We formulate the task as a structured prediction problem of assigning an object/background label to each pixel, implemented using a deep fully convolutional network. When applied to a video, our model further incorporates a motion stream, and the network learns to combine both appearance and motion and attempts to extract all prominent objects whether they are moving or not. Beyond the core model, a second contribution of our approach is how it leverages varying strengths of training annotations. Pixel-level annotations are quite difficult to obtain, yet crucial for training a deep network approach for segmentation. Thus we propose ways to exploit weakly labeled data for learning dense foreground segmentation. For images, we show the value in mixing object category examples with image-level labels together with relatively few images with boundary-level annotations. For video, we show how to bootstrap weakly annotated videos together with the network trained for image segmentation. Through experiments on multiple challenging image and video segmentation benchmarks, our method offers consistently strong results and improves the state-of-the-art for fully automatic segmentation of generic (unseen) objects. In addition, we demonstrate how our approach benefits image retrieval and image retargeting, both of which flourish when given our high-quality foreground maps. Code, models, and videos are at:http://vision.cs.utexas.edu/projects/pixelobjectness/
Unsupervised learning poses one of the most difficult challenges in computer vision today. The task has an immense practical value with many applications in artificial intelligence and emerging technologies, as large quantities of unlabeled videos can be collected at relatively low cost. In this paper, we address the unsupervised learning problem in the context of detecting the main foreground objects in single images. We train a student deep network to predict the output of a teacher pathway that performs unsupervised object discovery in videos or large image collections. Our approach is different from published methods on unsupervised object discovery. We move the unsupervised learning phase during training time, then at test time we apply the standard feed-forward processing along the student pathway. This strategy has the benefit of allowing increased generalization possibilities during training, while remaining fast at testing. Our unsupervised learning algorithm can run over several generations of student-teacher training. Thus, a group of student networks trained in the first generation collectively create the teacher at the next generation. In experiments our method achieves top results on three current datasets for object discovery in video, unsupervised image segmentation and saliency detection. At test time the proposed system is fast, being one to two orders of magnitude faster than published unsupervised methods.
The graph Laplacian is a standard tool in data science, machine learning, and image processing. The corresponding matrix inherits the complex structure of the underlying network and is in certain applications densely populated. This makes computations, in particular matrix-vector products, with the graph Laplacian a hard task. A typical application is the computation of a number of its eigenvalues and eigenvectors. Standard methods become infeasible as the number of nodes in the graph is too large. We propose the use of the fast summation based on the nonequispaced fast Fourier transform (NFFT) to perform the dense matrix-vector product with the graph Laplacian fast without ever forming the whole matrix. The enormous flexibility of the NFFT algorithm allows us to embed the accelerated multiplication into Lanczos-based eigenvalues routines or iterative linear system solvers and even consider other than the standard Gaussian kernels. We illustrate the feasibility of our approach on a number of test problems from image segmentation to semi-supervised learning based on graph-based PDEs. In particular, we compare our approach with the Nystr\"om method. Moreover, we present and test an enhanced, hybrid version of the Nystr\"om method, which internally uses the NFFT.
The interactive image segmentation algorithm can provide an intelligent ways to understand the intention of user input. Many interactive methods have the problem of that ask for large number of user input. To efficient produce intuitive segmentation under limited user input is important for industrial application. In this paper, we reveal a positive feedback system on image segmentation to show the pixels of self-learning. Two approaches, iterative random walks and boundary random walks, are proposed for segmentation potential, which is the key step in feedback system. Experiment results on image segmentation indicates that proposed algorithms can obtain more efficient input to random walks. And higher segmentation performance can be obtained by applying the iterative boundary random walks algorithm.
Segmentation of the left atrial chamber and assessing its morphology, are essential for improving our understanding of atrial fibrillation, the most common type of cardiac arrhythmia. Automation of this process in 3D gadolinium enhanced-MRI (GE-MRI) data is desirable, as manual delineation is time-consuming, challenging and observer-dependent. Recently, deep convolutional neural networks (CNNs) have gained tremendous traction and achieved state-of-the-art results in medical image segmentation. However, it is difficult to incorporate local and global information without using contracting (pooling) layers, which in turn reduces segmentation accuracy for smaller structures. In this paper, we propose a 3D CNN for volumetric segmentation of the left atrial chamber in LGE-MRI. Our network is based on the well known U-Net architecture. We employ a 3D fully convolutional network, with dilated convolutions in the lowest level of the network, and residual connections between encoder blocks to incorporate local and global knowledge. The results show that including global context through the use of dilated convolutions, helps in domain adaptation, and the overall segmentation accuracy is improved in comparison to a 3D U-Net.
Objective: Herein, a neural network-based liver segmentation algorithm is proposed, and its performance was evaluated using abdominal computed tomography (CT) images. Methods: A fully convolutional network was developed to overcome the volumetric image segmentation problem. To guide a neural network to accurately delineate a target liver object, the network was deeply supervised by applying the adaptive self-supervision scheme to derive the essential contour, which acted as a complement with the global shape. The discriminative contour, shape, and deep features were internally merged for the segmentation results. Results and Conclusion: 160 abdominal CT images were used for training and validation. The quantitative evaluation of the proposed network was performed through an eight-fold cross-validation. The result showed that the method, which uses the contour feature, segmented the liver more accurately than the state-of-the-art with a 2.13% improvement in the dice score. Significance: In this study, a new framework was introduced to guide a neural network and learn complementary contour features. The proposed neural network demonstrates that the guided contour features can significantly improve the performance of the segmentation task.
Confusing classes that are ubiquitous in real world often degrade performance for many vision related applications like object detection, classification, and segmentation. The confusion errors are not only caused by similar visual patterns but also amplified by various factors during the training of our designed models, such as reduced feature resolution in the encoding process or imbalanced data distributions. A large amount of deep learning based network structures has been proposed in recent years to deal with these individual factors and improve network performance. However, to our knowledge, no existing work in semantic image segmentation is designed to tackle confusion errors explicitly. In this paper, we present a novel and general network structure that reduces confusion errors in more direct manner and apply the network for semantic segmentation. There are two major contributions in our network structure: 1) We ensemble subnets with heterogeneous output spaces based on the discriminative confusing groups. The training for each subnet can distinguish confusing classes within the group without affecting unrelated classes outside the group. 2) We propose an improved cross-entropy loss function that maximizes the probability assigned to the correct class and penalizes the probabilities assigned to the confusing classes at the same time. Our network structure is a general structure and can be easily adapted to any other networks to further reduce confusion errors. Without any changes in the feature encoder and post-processing steps, our experiments demonstrate consistent and significant improvements on different baseline models on Cityscapes and PASCAL VOC datasets (e.g., 3.05% over ResNet-101 and 1.30% over ResNet-38).
In two-phase image segmentation, convex relaxation has allowed global minimisers to be computed for a variety of data fitting terms. Many efficient approaches exist to compute a solution quickly. However, we consider whether the nature of the data fitting in this formulation allows for reasonable assumptions to be made about the solution that can improve the computational performance further. In particular, we employ a well known dual formulation of this problem and solve the corresponding equations in a restricted domain. We present experimental results that explore the dependence of the solution on this restriction and quantify imrovements in the computational performance. This approach can be extended to analogous methods simply and could provide an efficient alternative for problems of this type.
Thalamic alterations are relevant to many neurological disorders including Alzheimer's disease, Parkinson's disease and multiple sclerosis. Routine interventions to improve symptom severity in movement disorders, for example, often consist of surgery or deep brain stimulation to diencephalic nuclei. Therefore, accurate delineation of grey matter thalamic subregions is of the upmost clinical importance. MRI is highly appropriate for structural segmentation as it provides different views of the anatomy from a single scanning session. Though with several contrasts potentially available, it is also of increasing importance to develop new image segmentation techniques that can operate multi-spectrally. We hereby propose a new segmentation method for use with multi-modality data, which we evaluated for automated segmentation of major thalamic subnuclear groups using T1-, T2*-weighted and quantitative susceptibility mapping (QSM) information. The proposed method consists of four steps: highly iterative image co-registration, manual segmentation on the average training-data template, supervised learning for pattern recognition, and a final convex optimisation step imposing further spatial constraints to refine the solution. This led to solutions in greater agreement with manual segmentation than the standard Morel atlas based approach. Furthermore, we show that the multi-contrast approach boosts segmentation performances. We then investigated whether prior knowledge using the training-template contours could further improve convex segmentation accuracy and robustness, which led to highly precise multi-contrast segmentations in single subjects. This approach can be extended to most 3D imaging data types and any region of interest discernible in single scans or multi-subject templates.