ABSTRACT
Disease in the brain is often associated with subtle, spatially diffuse, or complex tissue changes that may lie beneath the level of gross visual inspection, even on magnetic resonance imaging (MRI). Unfortunately, current computer-assisted approaches that examine pre-specified features, whether anatomically-defined (i.e. thalamic volume, cortical thickness) or based on pixelwise comparison (i.e. deformation-based methods), are prone to missing a vast array of physical changes that are not well-encapsulated by these metrics. In this paper, we have developed a technique for automated pattern analysis that can fully determine the relationship between brain structure and observable phenotype without requiring any a priori features. Our technique, called transport-based morphometry (TBM), is an image transformation that maps brain images losslessly to a domain where they become much more separable. The new approach is validated on structural brain images of healthy older adult subjects where even linear models for discrimination, regression, and blind source separation enable TBM to independently discover the characteristic changes of aging and highlight potential mechanisms by which aerobic fitness may mediate brain health later in life. TBM is a generative approach that can provide visualization of physically meaningful shifts in tissue distribution through inverse transformation. The proposed framework is a powerful technique that can potentially elucidate genotype-structural-behavioral associations in myriad diseases.
Subject(s)
Aging , Brain/diagnostic imaging , Image Interpretation, Computer-Assisted/methods , Image Processing, Computer-Assisted/methods , Magnetic Resonance Imaging/methods , Models, Theoretical , Pattern Recognition, Automated/methods , Aged , Biomarkers , HumansABSTRACT
Modern microscopic imaging devices are able to extract more information regarding the subcellular organization of different molecules and proteins than can be obtained by visual inspection. Predetermined numerical features (descriptors) often used to quantify cells extracted from these images have long been shown useful for discriminating cell populations (e.g., normal vs. diseased). Direct visual or biological interpretation of results obtained, however, is often not a trivial task. We describe an approach for detecting and visualizing phenotypic differences between classes of cells based on the theory of optimal mass transport. The method is completely automated, does not require the use of predefined numerical features, and at the same time allows for easily interpretable visualizations of the most significant differences. Using this method, we demonstrate that the distribution pattern of peripheral chromatin in the nuclei of cells extracted from liver and thyroid specimens is associated with malignancy. We also show the method can correctly recover biologically interpretable and statistically significant differences in translocation imaging assays in a completely automated fashion.
Subject(s)
Cells/cytology , Image Processing, Computer-Assisted/methods , Microscopy/methods , Molecular Imaging/methods , Phenotype , Algorithms , Humans , Microscopy/trends , Pattern Recognition, Automated/methods , Principal Component Analysis/methodsABSTRACT
Transport-based techniques for signal and data analysis have received increased attention recently. Given their ability to provide accurate generative models for signal intensities and other data distributions, they have been used in a variety of applications including content-based retrieval, cancer detection, image super-resolution, and statistical machine learning, to name a few, and shown to produce state of the art results in several applications. Moreover, the geometric characteristics of transport-related metrics have inspired new kinds of algorithms for interpreting the meaning of data distributions. Here we provide a practical overview of the mathematical underpinnings of mass transport-related methods, including numerical implementation, as well as a review, with demonstrations, of several applications. Software accompanying this tutorial is available at [43].
ABSTRACT
We present a new approach to facilitate the application of the optimal transport metric to pattern recognition on image databases. The method is based on a linearized version of the optimal transport metric, which provides a linear embedding for the images. Hence, it enables shape and appearance modeling using linear geometric analysis techniques in the embedded space. In contrast to previous work, we use Monge's formulation of the optimal transport problem, which allows for reasonably fast computation of the linearized optimal transport embedding for large images. We demonstrate the application of the method to recover and visualize meaningful variations in a supervised-learning setting on several image datasets, including chromatin distribution in the nuclei of cells, galaxy morphologies, facial expressions, and bird species identification. We show that the new approach allows for high-resolution construction of modes of variations and discrimination and can enhance classification accuracy in a variety of image discrimination problems.
ABSTRACT
Mesothelioma is a form of cancer generally caused from previous exposure to asbestos. Although it was considered a rare neoplasm in the past, its incidence is increasing worldwide due to extensive use of asbestos. In the current practice of medicine, the gold standard for diagnosing mesothelioma is through a pleural biopsy with subsequent histologic examination of the tissue. The diagnostic tissue should demonstrate the invasion by the tumor and is obtained through thoracoscopy or open thoracotomy, both being highly invasive surgical operations. On the other hand, thoracocentesis, which is removal of effusion fluid from the pleural space, is a far less invasive procedure that can provide material for cytological examination. In this study, we aim at detecting and classifying malignant mesothelioma based on the nuclear chromatin distribution from digital images of mesothelial cells in effusion cytology specimens. Accordingly, a computerized method is developed to determine whether a set of nuclei belonging to a patient is benign or malignant. The quantification of chromatin distribution is performed by using the optimal transport-based linear embedding for segmented nuclei in combination with the modified Fisher discriminant analysis. Classification is then performed through a k-nearest neighborhood approach and a basic voting strategy. Our experiments on 34 different human cases result in 100% accurate predictions computed with blind cross validation. Experimental comparisons also show that the new method can significantly outperform standard numerical feature-type methods in terms of agreement with the clinical diagnosis gold standard. According to our results, we conclude that nuclear structure of mesothelial cells alone may contain enough information to separate malignant mesothelioma from benign mesothelial proliferations.
Subject(s)
Cell Nucleus/physiology , Cytodiagnosis/methods , Lung Neoplasms/classification , Lung Neoplasms/diagnosis , Mesothelioma/classification , Mesothelioma/diagnosis , Pleural Effusion, Malignant/cytology , Asbestos/adverse effects , Chromatin/physiology , Cytological Techniques/methods , Epithelial Cells/pathology , Humans , Image Processing, Computer-Assisted , Mesothelioma, Malignant , Pleura/cytology , Pleura/pathology , Pleural Effusion, Malignant/pathologyABSTRACT
Acoustic travel-time tomography of the atmosphere is a nonlinear inverse problem which attempts to reconstruct temperature and wind velocity fields in the atmospheric surface layer using the dependence of sound speed on temperature and wind velocity fields along the propagation path. This paper presents a statistical-based acoustic travel-time tomography algorithm based on dual state-parameter unscented Kalman filter (UKF) which is capable of reconstructing and tracking, in time, temperature, and wind velocity fields (state variables) as well as the dynamic model parameters within a specified investigation area. An adaptive 3-D spatial-temporal autoregressive model is used to capture the state evolution in the UKF. The observations used in the dual state-parameter UKF process consist of the acoustic time of arrivals measured for every pair of transmitter/receiver nodes deployed in the investigation area. The proposed method is then applied to the data set collected at the Meteorological Observatory Lindenberg, Germany, as part of the STINHO experiment, and the reconstruction results are presented.
Subject(s)
Acoustics , Atmosphere , Models, Statistical , Signal Processing, Computer-Assisted , Sound , Algorithms , Motion , Sound Spectrography , Temperature , Time Factors , WindABSTRACT
Blue light cystoscopy (BLC) is a guideline-recommended endoscopic tool to detect bladder cancer with high sensitivity. Having clear, high-quality images during cystoscopy is crucial to the sensitive, efficient detection of bladder tumors; yet, important diagnostic information is often missed or poorly visualized in images containing illumination artifacts or impacted by impurities in the bladder. In this study, we introduce computational methods to remove two common artifacts in images from BLC videos: green hue and fogginess. We also evaluate the effect of artifact removal on the perceptual quality of the BLC images through a survey study and computation of Blind/Referenceless Image Spatial Quality Evaluator scores on the original and enhanced images. We show that corrections and enhancements made to cystoscopy images resulted in a better viewing experience for clinicians during BLC imaging and reliably restored lost tissue features that were important for diagnostics. Incorporating these enhancements during clinical and OR procedures may lead to more comprehensive tumor detection, fewer missed tumors during TURBT procedures, more complete tumor resection and shorter procedure time. When used in off-line review of cystoscopy videos, it may also better guide surgical planning and allow more accurate assessment and diagnosis.
Subject(s)
Aminolevulinic Acid , Urinary Bladder Neoplasms , Humans , Cystoscopy/methods , Urinary Bladder Neoplasms/diagnostic imaging , Urinary Bladder Neoplasms/pathology , CystectomyABSTRACT
Despite the advancement of machine learning techniques in recent years, state-of-the-art systems lack robustness to "real world" events, where the input distributions and tasks encountered by the deployed systems will not be limited to the original training context, and systems will instead need to adapt to novel distributions and tasks while deployed. This critical gap may be addressed through the development of "Lifelong Learning" systems that are capable of (1) Continuous Learning, (2) Transfer and Adaptation, and (3) Scalability. Unfortunately, efforts to improve these capabilities are typically treated as distinct areas of research that are assessed independently, without regard to the impact of each separate capability on other aspects of the system. We instead propose a holistic approach, using a suite of metrics and an evaluation framework to assess Lifelong Learning in a principled way that is agnostic to specific domains or system techniques. Through five case studies, we show that this suite of metrics can inform the development of varied and complex Lifelong Learning systems. We highlight how the proposed suite of metrics quantifies performance trade-offs present during Lifelong Learning system development - both the widely discussed Stability-Plasticity dilemma and the newly proposed relationship between Sample Efficient and Robust Learning. Further, we make recommendations for the formulation and use of metrics to guide the continuing development of Lifelong Learning systems and assess their progress in the future.
Subject(s)
Education, Continuing , Machine LearningABSTRACT
In this article, we consider a subclass of partially observable Markov decision process (POMDP) problems which we termed confounding POMDPs. In these types of POMDPs, temporal difference (TD)-based reinforcement learning (RL) algorithms struggle, as TD error cannot be easily derived from observations. We solve these types of problems using a new bio-inspired neural architecture that combines a modulated Hebbian network (MOHN) with deep Q-network (DQN), which we call modulated Hebbian plus Q-network architecture (MOHQA). The key idea is to use a Hebbian network with rarely correlated bio-inspired neural traces to bridge temporal delays between actions and rewards when confounding observations and sparse rewards result in inaccurate TD errors. In MOHQA, DQN learns low-level features and control, while the MOHN contributes to high-level decisions by associating rewards with past states and actions. Thus, the proposed architecture combines two modules with significantly different learning algorithms, a Hebbian associative network and a classical DQN pipeline, exploiting the advantages of both. Simulations on a set of POMDPs and on the Malmo environment show that the proposed algorithm improved DQN's results and even outperformed control tests with advantage-actor critic (A2C), quantile regression DQN with long short-term memory (QRDQN + LSTM), Monte Carlo policy gradient (REINFORCE), and aggregated memory for reinforcement learning (AMRL) algorithms on most difficult POMDPs with confounding stimuli and sparse rewards.
Subject(s)
Neural Networks, Computer , Reinforcement, Psychology , Algorithms , Markov Chains , RewardABSTRACT
We present a new supervised image classification method applicable to a broad class of image deformation models. The method makes use of the previously described Radon Cumulative Distribution Transform (R-CDT) for image data, whose mathematical properties are exploited to express the image data in a form that is more suitable for machine learning. While certain operations such as translation, scaling, and higher-order transformations are challenging to model in native image space, we show the R-CDT can capture some of these variations and thus render the associated image classification problems easier to solve. The method - utilizing a nearest-subspace algorithm in the R-CDT space - is simple to implement, non-iterative, has no hyper-parameters to tune, is computationally efficient, label efficient, and provides competitive accuracies to state-of-the-art neural networks for many types of classification problems. In addition to the test accuracy performances, we show improvements (with respect to neural network-based methods) in terms of computational efficiency (it can be implemented without the use of GPUs), number of training samples needed for training, as well as out-of-distribution generalization. The Python code for reproducing our results is available at [1].
ABSTRACT
In uncertain domains, the goals are often unknown and need to be predicted by the organism or system. In this paper, contrastive Excitation Backprop (c-EB) was used in two goal-driven perception tasks - one with pairs of noisy MNIST digits and the other with a robot in an action-based attention scenario. The first task included attending to even, odd, low, and high digits, whereas the second task included action goals, such as "eat", "work-on-computer", "read", and "say-hi" that led to attention to objects associated with those actions. The system needed to increase attention to target items and decrease attention to distractor items and background noise. Because the valid goal was unknown, an online learning model based on the cholinergic and noradrenergic neuromodulatory systems was used to predict a noisy goal (expected uncertainty) and re-adapt when the goal changed (unexpected uncertainty). This neurobiologically plausible model demonstrates how neuromodulatory systems can predict goals in uncertain domains and how attentional mechanisms can enhance the perception for that goal.
Subject(s)
Attention , Models, Neurological , Neural Networks, Computer , Uncertainty , Goals , Humans , Perception , Reaction TimeABSTRACT
The ability of an agent to detect changes in an environment is key to successful adaptation. This ability involves at least two phases: learning a model of an environment, and detecting that a change is likely to have occurred when this model is no longer accurate. This task is particularly challenging in partially observable environments, such as those modeled with partially observable Markov decision processes (POMDPs). Some predictive learners are able to infer the state from observations and thus perform better with partial observability. Predictive state representations (PSRs) and neural networks are two such tools that can be trained to predict the probabilities of future observations. However, most such existing methods focus primarily on static problems in which only one environment is learned. In this paper, we propose an algorithm that uses statistical tests to estimate the probability of different predictive models to fit the current environment. We exploit the underlying probability distributions of predictive models to provide a fast and explainable method to assess and justify the model's beliefs about the current environment. Crucially, by doing so, the method can label incoming data as fitting different models, and thus can continuously train separate models in different environments. This new method is shown to prevent catastrophic forgetting when new environments, or tasks, are encountered. The method can also be of use when AI-informed decisions require justifications because its beliefs are based on statistical evidence from observations. We empirically demonstrate the benefit of the novel method with simulations in a set of POMDP environments.
ABSTRACT
Transport based distances, such as the Wasserstein distance and earth mover'sdistance, have been shown to be an effective tool in signal and image analysis. The success of transport based distances is in part due to their Lagrangian nature which allows it to capture the important variations in many signal classes. However these distances require the signal to be nonnegative and normalized. Furthermore, the signals are considered as measures and compared by redistributing (transporting) them, which does not directly take into account the signal intensity. Here we study a transport-based distance, called the TLp distance, that combines Lagrangian and intensity modelling and is directly applicable to general, non-positive and multi-channelled signals. The distance can be computed by existing numerical methods. We give an overview of the basic properties of this distance and applications to classification, with multi-channelled non-positive one-dimensional signals and two-dimensional images, and color transfer.
ABSTRACT
Invertible image representation methods (transforms) are routinely employed as low-level image processing operations based on which feature extraction and recognition algorithms are developed. Most transforms in current use (e.g., Fourier, wavelet, and so on) are linear transforms and, by themselves, are unable to substantially simplify the representation of image classes for classification. Here, we describe a nonlinear, invertible, low-level image processing transform based on combining the well-known Radon transform for image data, and the 1D cumulative distribution transform proposed earlier. We describe a few of the properties of this new transform, and with both theoretical and experimental results show that it can often render certain problems linearly separable in a transform space.
Subject(s)
Algorithms , Image Processing, Computer-Assisted/methods , Signal Processing, Computer-Assisted , Face/anatomy & histology , Humans , Pattern Recognition, Automated/methodsABSTRACT
OBJECTIVES: Evaluate the effects of (1) laryngeal configuration, (2) image barrel distortion, and (3) lens perspective on the accuracy of anterior and posterolateral laryngeal lumen angle morphometric estimates. STUDY DESIGN: Prospective, repeated-measures. METHODS: Twenty-four (24) excised canine larynges were manipulated across four different length-by-width configurations to evaluate the influence of laryngeal configuration on relative precision of three laryngeal morphometric angle estimate methods. Physical measurements of the laryngeal specimens were compared (statistically and descriptively) with corresponding unadjusted and barrel-distortion corrected laryngoscopic laryngeal images. Additional post hoc analysis involved systematic manipulation of camera lens-to-object perspective (shift, tilt, and distance) using a synthetic object representing the physical laryngeal specimen and simulated image of the laryngoscopic laryngeal images. Morphometric angle estimates between the synthetic object and the simulated images were compared across simulated lens-to-object perspective manipulations to evaluate influences of lens perspective artifacts on laryngeal morphometric estimate precision. RESULTS: Statistical analysis showed that laryngeal morphometric angle estimates were impervious to laryngeal configuration manipulations, but were influenced by image barrel-correction methods. Statistically significant differences were found between the unadjusted and barrel-corrected images within the anterior angle method. Simulated camera lens-to-object perspective manipulations showed that tilt and distance have substantial negative influence on laryngeal morphometric estimate precision. CONCLUSIONS: Laryngeal lumen angles can be used to measure respiratory laryngeal morphometry. However, image-correction algorithms are necessary to correct images and quantify morphometric estimate error caused by camera lens distortion and lens-to-object perspective. Findings provide a platform for future research on quantifying laryngeal morphometry, especially for individuals with laryngeal breathing disorders.
Subject(s)
Image Processing, Computer-Assisted/methods , Laryngoscopy/methods , Larynx/anatomy & histology , Larynx/physiology , Optics and Photonics/methods , Respiration , Algorithms , Animals , Artifacts , Biomechanical Phenomena , Dogs , Equipment Design , Laryngoscopes , Laryngoscopy/instrumentation , Lenses , Models, Animal , Optics and Photonics/instrumentation , Reproducibility of ResultsABSTRACT
Follicular lesions of the thyroid remain significant diagnostic challenges in surgical pathology and cytology. The diagnosis often requires considerable resources and ancillary tests including immunohistochemistry, molecular studies, and expert consultation. Visual analyses of nuclear morphological features, generally speaking, have not been helpful in distinguishing this group of lesions. Here we describe a method for distinguishing between follicular lesions of the thyroid based on nuclear morphology. The method utilizes an optimal transport-based linear embedding for segmented nuclei, together with an adaptation of existing classification methods. We show the method outputs assignments (classification results) which are near perfectly correlated with the clinical diagnosis of several lesion types' lesions utilizing a database of 94 patients in total. Experimental comparisons also show the new method can significantly outperform standard numerical feature-type methods in terms of agreement with the clinical diagnosis gold standard. In addition, the new method could potentially be used to derive insights into biologically meaningful nuclear morphology differences in these lesions. Our methods could be incorporated into a tool for pathologists to aid in distinguishing between follicular lesions of the thyroid. In addition, these results could potentially provide nuclear morphological correlates of biological behavior and reduce health care costs by decreasing histotechnician and pathologist time and obviating the need for ancillary testing.