Image Analysis

Effective convolutional neural networks are trained on large sets of labeled data. However, creating large labeled datasets is a very costly and time-consuming task. Semi-supervised learning uses unlabeled data to train a model with higher accuracy when there is a limited set of labeled data available. In this paper, we consider the problem of semi-supervised learning with convolutional neural networks. Techniques such as randomized data augmentation, dropout and random max-pooling provide better generalization and stability for classifiers that are trained using gradient descent. Multiple passes of an individual sample through the network might lead to different predictions due to the non-deterministic behavior of these techniques. We propose an unsupervised loss function that takes advantage of the stochastic nature of these methods and minimizes the difference between the predictions of multiple passes of a training sample through the network. We evaluate the proposed method on several benchmark datasets.

Artificial neural networks are powerful pattern classifiers. They form the basis of the highly successful and popular Convolutional Networks which offer the state-of-the-art performance on several computer visions tasks. However, in many general and non-vision tasks, neural networks are surpassed by methods such as support vector machines and random forests that are also easier to use and faster to train. One reason is that the backpropagation algorithm, which is used to train artificial neural networks, usually starts from a random weight initialization which complicates the optimization process leading to long training times and increases the risk of stopping in a poor local minima. Several initialization schemes and pre-training methods have been proposed to improve the efficiency and performance of training a neural network. However, this problem arises from the architecture of neural networks. We use the disjunctive normal form and approximate the boolean conjunction operations with products to construct a novel network architecture. The proposed model can be trained by minimizing an error function and it allows an effective and intuitive initialization which avoids poor local minima. We show that the proposed structure provides efficient coverage of the decision space which leads to state-of-the art classification accuracy and fast training times.

This paper investigates one of the most fundamental computer vision problems: image segmentation. We propose a supervised hierarchical approach to object-independent image segmentation. Starting with oversegmenting superpixels, we use a tree structure to represent the hierarchy of region merging, by which we reduce the problem of segmenting image regions to finding a set of label assignment to tree nodes. We formulate the tree structure as a constrained conditional model to associate region merging with likelihoods predicted using an ensemble boundary classifier. Final segmentations can then be inferred by finding globally optimal solutions to the model efficiently. We also present an iterative training and testing algorithm that generates various tree structures and combines them to emphasize accurate boundaries by segmentation accumulation. Experiment results and comparisons with other recent methods on six public datasets demonstrate that our approach achieves the state-of-the-art region accuracy and is competitive in image segmentation without semantic priors.

Current late gadolinium enhancement (LGE) imaging of left atrial (LA) scar or fibrosis is relatively slow and requires 5–15 min to acquire an undersampled (R = 1.7) 3D navigated dataset. The GeneRalized Autocalibrating Partially Parallel Acquisitions (GRAPPA) based parallel imaging method is the current clinical standard for accelerating 3D LGE imaging of the LA and permits an acceleration factor ~ R = 1.7. Two compressed sensing (CS) methods have been developed to achieve higher acceleration factors: a patch based collaborative filtering technique tested with acceleration factor R ~ 3, and a technique that uses a 3D radial stack-of-stars acquisition pattern (R ~ 1.8) with a 3D total variation constraint. The long reconstruction time of these CS methods makes them unwieldy to use, especially the patch based collaborative filtering technique. In addition, the effect of CS techniques on the quantification of percentage of scar/fibrosis is not known.

We sought to develop a practical compressed sensing method for imaging the LA at high acceleration factors. In order to develop a clinically viable method with short reconstruction time, a Split Bregman (SB) reconstruction method with 3D total variation (TV) constraints was developed and implemented. The method was tested on 8 atrial fibrillation patients (4 pre-ablation and 4 post-ablation datasets). Blur metric, normalized mean squared error and peak signal to noise ratio were used as metrics to analyze the quality of the reconstructed images, Quantification of the extent of LGE was performed on the undersampled images and compared with the fully sampled images. Quantification of scar from post-ablation datasets and quantification of fibrosis from pre-ablation datasets showed that acceleration factors up to R ~ 3.5 gave good 3D LGE images of the LA wall, using a 3D TV constraint and constrained SB methods. This corresponds to reducing the scan time by half, compared to currently used GRAPPA methods. Reconstruction of 3D LGE images using the SB method was over 20 times faster than standard gradient descent methods.

Purpose:
Rapid reconstruction of undersampled multicoil MRI data with iterative constrained reconstruction method is a challenge. The authors sought to develop a new substitution based variable splitting algorithm for faster reconstruction of multicoil cardiac perfusion MRI data.

Methods:
The new method, split Bregman multicoil accelerated reconstruction technique (SMART), uses a combination of split Bregman based variable splitting and iterative reweighting techniques to achieve fast convergence. Total variation constraints are used along the spatial and temporal dimensions. The method is tested on nine ECG-gated dog perfusion datasets, acquired with a 30-ray golden ratio radial sampling pattern and ten ungated human perfusion datasets, acquired with a 24-ray golden ratio radial sampling pattern. Image quality and reconstruction speed are evaluated and compared to a gradient descent (GD) implementation and to multicoil k-t SLR, a reconstruction technique that uses a combination of sparsity and low rank constraints.

Results:
Comparisons based on blur metric and visual inspection showed that SMART images had lower blur and better texture as compared to the GD implementation. On average, the GD based images had an ∼18% higher blur metric as compared to SMART images. Reconstruction of dynamic contrast enhanced (DCE) cardiac perfusion images using the SMART method was ∼6 times faster than standard gradient descent methods. k-t SLR and SMART produced images with comparable image quality, though SMART was ∼6.8 times faster than k-t SLR.

Conclusions:
The SMART method is a promising approach to reconstruct good quality multicoil images from undersampled DCE cardiac perfusion data rapidly.

Semantic segmentation is the problem of assigning an object label to each pixel. It unifies the image segmentation and object recognition problems. The importance of using contextual information in semantic segmentation frameworks has been widely realized in the field. We propose a contextual framework, called contextual hierarchical model (CHM), which learns contextual information in a hierarchical framework for semantic segmentation. At each level of the hierarchy, a classifier is trained based on downsampled input images and outputs of previous levels. Our model then incorporates the resulting multi-resolution contextual information into a classifier to segment the input image at original resolution. This training strategy allows for optimization of a joint posterior probability at multiple resolutions through the hierarchy. Contextual hierarchical model is purely based on the input image patches and does not make use of any fragments or shape examples. Hence, it is applicable to a variety of problems such as object segmentation and edge detection. We demonstrate that CHM performs at par with state-of-the-art on Stanford background and Weizmann horse datasets. It also outperforms state-of-the-art edge detection methods on NYU depth dataset and achieves state-of-the-art on Berkeley segmentation dataset (BSDS 500).

The recent precipitous losses of summer Arctic sea ice have outpaced the projections of most climate models. A number of efforts to improve these models have focused in part on a more accurate accounting of sea ice albedo or reflectance. In late spring and summer, the albedo of the ice pack is determined primarily by melt ponds that form on the sea ice surface. The transition of pond configurations from isolated structures to interconnected networks is critical in allowing the lateral flow of melt water toward drainage features such as large brine channels, fractures, and seal holes, which can alter the albedo by removing the melt water. Moreover, highly connected ponds can influence the formation of fractures and leads during ice break-up. Here we develop algorithmic techniques for mapping photographic images of melt ponds onto discrete conductance networks which represent the geometry and connectedness of pond configurations. The effective conductivity of the networks is computed to approximate the ease of lateral flow. We implement an image processing algorithm with mathematical morphology operations to produce a conductance matrix representation of the melt ponds. Basic clustering and edge elimination, using undirected graphs, are then used to map the melt pond connections and reduce the conductance matrix to include only direct connections. The results for images taken during different times of the year are visually inspected and the number of mislabels is used to evaluate performance.

The visualization of variability in surfaces embedded in 3D, which is a type of ensemble uncertainty visualization, provides a means of understanding the underlying distribution of a collection or ensemble of surfaces. This work extends the contour boxplot technique to 3D and evaluates it against an enumeration-style visualization of the ensemble members and other conventional visualizations used by atlas builders. The authors demonstrate the efficacy of using the 3D contour boxplot ensemble visualization technique to analyze shape alignment and variability in atlas construction and analysis as a real-world application.

The quantification of local surface morphology in the human cortex is important for examining population differences as well as developmental changes in neurodegenerative or neurodevelopmental disorders. We propose a novel cortical shape measure, referred to as the 'shape complexity index' (SCI), that represents localized shape complexity as the difference between the observed distributions of local surface topology, as quantified by the shape index (SI) measure, to its best fitting simple topological model within a given neighborhood. We apply a relatively small, adaptive geodesic kernel to calculate the SCI. Due to the small size of the kernel, the proposed SCI measure captures fine differences of cortical shape. With this novel cortical feature, we aim to capture comparatively small local surface changes that capture a) the widening versus deepening of sulcal and gyral regions, as well as b) the emergence and development of secondary and tertiary sulci. Current cortical shape measures, such as the gyrification index (GI) or intrinsic curvature measures, investigate the cortical surface at a different scale and are less well suited to capture these particular cortical surface changes. In our experiments, the proposed SCI demonstrates higher complexity in the gyral/sulcal wall regions, lower complexity in wider gyral ridges and lowest complexity in wider sulcal fundus regions. In early postnatal brain development, our experiments show that SCI reveals a pattern of increased cortical shape complexity with age, as well as sexual dimorphisms in the insula, middle cingulate, parieto-occipital sulcal and Broca's regions. Overall, sex differences were greatest at 6months of age and were reduced at 24months, with the difference pattern switching from higher complexity in males at 6months to higher complexity in females at 24months. This is the first study of longitudinal, cortical complexity maturation and sex differences, in the early postnatal period from 6 to 24months of age with fine scale, cortical shape measures. These results provide information that complement previous studies of gyrification index in early brain development.

Probabilistic label maps are a useful tool for important medical image analysis tasks such as segmentation, shape analysis, and atlas building. Existing methods typically rely on blurred signed distance maps or smoothed label maps to model uncertainties and shape variabilities, which do not conform to any generative model or estimation process, and are therefore suboptimal. In this paper, we propose to learn probabilistic label maps using a generative model on given set of binary label maps. The proposed approach generalizes well on unseen data while simultaneously capturing the variability in the training samples. Efficiency of the proposed approach is demonstrated for consensus generation and shape-based clustering using synthetic datasets as well as left atrial segmentations from late-gadolinium enhancement MRI.

Diffusion-weighted (DW) MRI has become a widely adopted imaging modality to reveal the underlying brain connectivity. Long acquisition times and/or non-cooperative patients increase the chances of motion-related artifacts. Whereas slow bulk motion results in inter-gradient misalignment which can be handled via retrospective motion correction algorithms, fast bulk motion usually affects data during the application of a single diffusion gradient causing signal dropout artifacts. Common practices opt to discard gradients bearing signal attenuation due to the difficulty of their retrospective correction, with the disadvantage to lose full gradients for further processing. Nonetheless, such attenuation might only affect limited number of slices within a gradient volume. Q-space resampling has recently been proposed to recover corrupted slices while saving gradients for subsequent reconstruction. However, few corrupted gradients are implicitly assumed which might not hold in case of scanning unsedated infants or patients in pain. In this paper, we propose to adopt recent advances in compressive sensing based reconstruction of the diffusion orientation distribution functions (ODF) with under sampled measurements to resample corrupted slices. We make use of Simple Harmonic Oscillator based Reconstruction and Estimation (SHORE) basis functions which can analytically model ODF from arbitrary sampled signals. We demonstrate the impact of the proposed resampling strategy compared to state-of-art resampling and gradient exclusion on simulated intra-gradient motion as well as samples from real DWI data.

In this paper we present a novel method for analyzing the relationship between functional brain networks and behavioral phenotypes. Drawing from topological data analysis, we first extract topological features using persistent homology from functional brain networks that are derived from correlations in resting-state fMRI. Rather than fixing a discrete network topology by thresholding the connectivity matrix, these topological features capture the network organization across all continuous threshold values. We then propose to use a kernel partial least squares (kPLS) regression to statistically quantify the relationship between these topological features and behavior measures. The kPLS also provides an elegant way to combine multiple image features by using linear combinations of multiple kernels. In our experiments we test the ability of our proposed brain network analysis to predict autism severity from rs-fMRI. We show that combining correlations with topological features gives better prediction of autism severity than using correlations alone.

The goal of longitudinal shape analysis is to understand how anatomical shape changes over time, in response to biological processes, including growth, aging, or disease. In many imaging studies, it is also critical to understand how these shape changes are affected by other factors, such as sex, disease diagnosis, IQ, etc. Current approaches to longitudinal shape analysis have focused on modeling age-related shape changes, but have not included the ability to handle covariates. In this paper, we present a novel Bayesian mixed-effects shape model that incorporates simultaneous relationships between longitudinal shape data and multiple predictors or covariates to the model. Moreover, we place an Automatic Relevance Determination (ARD) prior on the parameters, that lets us automatically select which covariates are most relevant to the model based on observed data. We evaluate our proposed model and inference procedure on a longitudinal study of Huntington's disease from PREDICT-HD. We first show the utility of the ARD prior for model selection in a univariate modeling of striatal volume, and next we apply the full high-dimensional longitudinal shape model to putamen shapes.

A novel implicit parametric shape model is proposed for segmentation and analysis of medical images. Functions representing the shape of an object can be approximated as a union of N polytopes. Each polytope is obtained by the intersection of M half-spaces. The shape function can be approximated as a disjunction of conjunctions, using the disjunctive normal form. The shape model is initialized using seed points defined by the user. We define a cost function based on the Chan-Vese energy functional. The model is differentiable, hence, gradient based optimization algorithms are used to find the model parameters.

Shape priors have been successfully used in challenging biomedical imaging problems. However when the shape distribution involves multiple shape classes, leading to a multimodal shape density, effective use of shape priors in segmentation becomes more challenging. In such scenarios, knowing the class of the shape can aid the segmentation process, which is of course unknown a priori. In this paper, we propose a joint classification and segmentation approach for dendritic spine segmentation which infers the class of the spine during segmentation and adapts the remaining segmentation process accordingly. We evaluate our proposed approach on 2-photon microscopy images containing dendritic spines and compare its performance quantitatively to an existing approach based on nonparametric shape priors. Both visual and quantitative results demonstrate the effectiveness of our approach in dendritic spine segmentation.

Functional properties of a neuron are coupled with its morphology, particularly the morphology of dendritic spines. Spine volume has been used as the primary morphological parameter in order the characterize the structure and function coupling. However, this reductionist approach neglects the rich shape repertoire of dendritic spines. First step to incorporate spine shape information into functional coupling is classifying main spine shapes that were proposed in the literature. Due to the lack of reliable and fully automatic tools to analyze the morphology of the spines, such analysis is often performed manually, which is a laborious and time intensive task and prone to subjectivity. In this paper we present an automated approach to extract features using basic image processing techniques, and classify spines into mushroom or stubby by applying machine learning algorithms. Out of 50 manually segmented mushroom and stubby spines, Support Vector Machine was able to classify 98% of the spines correctly.

The use of appearance and shape priors in image segmentation is known to improve accuracy; however, existing techniques have several drawbacks. Active shape and appearance models require landmark points and assume unimodal shape and appearance distributions. Level set based shape priors are limited to global shape similarity. In this paper, we present a novel shape and appearance priors for image segmentation based on an implicit parametric shape representation called disjunctive normal shape model (DNSM). DNSM is formed by disjunction of conjunctions of half-spaces defined by discriminants. We learn shape and appearance statistics at varying spatial scales using nonparametric density estimation. Our method can generate a rich set of shape variations by locally combining training shapes. Additionally, by studying the intensity and texture statistics around each discriminant of our shape model, we construct a local appearance probability map. Experiments carried out on both medical and natural image datasets show the potential of the proposed method.

To stimulate progress in automating the reconstruction of neural circuits, we organized the first international challenge on 2D segmentation of electron microscopic (EM) images of the brain. Participants submitted boundary maps predicted for a test set of images, and were scored based on their agreement with a consensus of human expert annotations. The winning team had no prior experience with EM images, and employed a convolutional network. This "deep learning" approach has since become accepted as a standard for segmentation of EM images. The challenge has continued to accept submissions, and the best so far has resulted from cooperation between two teams. The challenge has probably saturated, as algorithms cannot progress beyond limits set by ambiguities inherent in 2D scoring and the size of the test dataset. Retrospective evaluation of the challenge scoring system reveals that it was not sufficiently robust to variations in the widths of neurite borders. We propose a solution to this problem, which should be useful for a future 3D segmentation challenge.

Background
In the area of connectomics, there is a significant gap between the time required for data acquisition and dense reconstruction of the neural processes contained in the same dataset. Automatic methods are able to eliminate this timing gap, but the state-of-the-art accuracy so far is insufficient for use without user corrections. If completed naively, this process of correction can be tedious and time consuming.

New method
We present a new semi-automatic method that can be used to perform 3D segmentation of neurites in EM image stacks. It utilizes an automatic method that creates a hierarchical structure for recommended merges of superpixels. The user is then guided through each predicted region to quickly identify errors and establish correct links.

Results
We tested our method on three datasets with both novice and expert users. Accuracy and timing were compared with published automatic, semi-automatic, and manual results.

Comparison with existing methods
Post-automatic correction methods have also been used in Mishchenko et al. (2010) and Haehn et al. (2014). These methods do not provide navigation or suggestions in the manner we present. Other semi-automatic methods require user input prior to the automatic segmentation such as Jeong et al. (2009) and Cardona et al. (2010) and are inherently different than our method.

Conclusion
Using this method on the three datasets, novice users achieved accuracy exceeding state-of-the-art automatic results, and expert users achieved accuracy on par with full manual labeling but with a 70% time improvement when compared with other examples in publication.