The application of machine learning and computer vision in microtomography provides new opportunities to directly analyze the microstructural evolutions of strain-hardening cementitious composites (SHCC) under tensile load, especially the strain-hardening process. For the first time, a state-of-the-art machine-learning pipeline combined with digital volume correlation for automated microtomography segmentation analysis (MSA) was developed to separate different components and quantify the in-situ 3D morphological properties of the fibers and pore networks imaged with in-situ synchrotron X-ray computed microtomography. Strain localization and crack initiation were observed around the interconnected pores where strain localized instead of the weakest cross-section defined by the fiber distribution and porosity. Fibers reinforced the crack planes through fiber debonding, bridging, bending, stretching, and orientation redistribution, which contributed to the crack width control and ductility of SHCC in the experiment. This work is essential to understand the progressive damage mechanisms of SHCC and help refine the characterization, modeling, and design of the composite using a bottom-up approach.

Abnormal tau inclusions are hallmarks of Alzheimer's disease and predictors of clinical decline. Several tau PET tracers are available for neurodegenerative disease research, opening avenues for molecular diagnosis in vivo. However, few have been approved for clinical use. Understanding the neurobiological basis of PET signal validation remains problematic because it requires a large-scale, voxel-to-voxel correlation between PET and (immuno) histological signals. Large dimensionality of whole human brains, tissue deformation impacting co-registration, and computing requirements to process terabytes of information preclude proper validation. We developed a computational pipeline to identify and segment particles of interest in billion-pixel digital pathology images to generate quantitative, 3D density maps. The proposed convolutional neural network for immunohistochemistry samples, IHCNet, is at the pipeline's core. We have successfully processed and immunostained over 500 slides from two whole human brains with three phospho-tau antibodies (AT100, AT8, and MC1), spanning several terabytes of images. Our artificial neural network estimated tau inclusion from brain images, which performs with ROC AUC of 0.87, 0.85, and 0.91 for AT100, AT8, and MC1, respectively. Introspection studies further assessed the ability of our trained model to learn tau-related features. We present an end-to-end pipeline to create terabytes-large 3D tau inclusion density maps co-registered to MRI as a means to facilitate validation of PET tracers.

A análise de células cervicais a partir de exames de Papanicolaou convencionais ainda é um grande desafio. Diferentemente das imagens de exame em meio líquido, a citologia convencional possui muita sobreposição celular e diversas estruturas epiteliais que dificultam a implementação de metodologias computacionais que possam dar suporte à automação dos exames. Este trabalho compara o desempenho de duas arquiteturas de redes neurais convolucionais para segmentar núcleos celulares. A avaliação considerou o uso de um novo banco de dados de segmentação. Os resultados mostraram que nossa proposta consegue segmentar núcleos celulares a partir de imagens de citologia convencional com múltiplas células e sobreposição.

Screening of Pap smear images continues to depend upon cytopathologists’ manual scrutiny, and the results are highly influenced by professional experience, leading to varying degrees of cell classification inaccuracies. In order to improve the quality of the Pap smear results, several efforts have been made to create software to automate and standardize the processing of medical images. In this work, we developed the CEA (Cytopathologist Eye Assistant), an easy-to-use tool to aid cytopathologists in performing their daily activities. In addition, the tool was tested by a group of cytopathologists, whose feedback indicates that CEA could be a valuable tool to be integrated into Pap smear image analysis routines. For the construction of the tool, we evaluate different YOLO configurations and classification approaches. The best combination of algorithms uses YOLOv5s as a detection algorithm and an ensemble of EfficientNets as a classification algorithm. This configuration achieved 0.726 precision, 0.906 recall, and 0.805 F1-score when considering individual cells. We also made an analysis to classify the image as a whole, in which case, the best configuration was the YOLOv5s to perform the detection and classification tasks, and it achieved 0.975 precision, 0.992 recall, 0.970 accuracy, and 0.983 F1-score.

The execution and analysis of complex experiments are challenged by the vast dimensionality of the underlying parameter spaces. Although an increase in data-acquisition rates should allow broader querying of the parameter space, the complexity of experiments and the subtle dependence of the model function on input parameters remains daunting owing to the sheer number of variables. New strategies for autonomous data acquisition are being developed, with one promising direction being the use of Gaussian process regression (GPR). GPR is a quick, non-parametric and robust approximation and uncertainty quantification method that can be applied directly to autonomous data acquisition. We review GPR-driven autonomous experimentation and illustrate its functionality using real-world examples from large experimental facilities in the USA and France. We introduce the basics of a GPR-driven autonomous loop with a focus on Gaussian processes, and then shift the focus to the infrastructure that needs to be built around GPR to create a closed loop. Finally, the case studies we discuss show that Gaussian-process-based autonomous data acquisition is a widely applicable method that can facilitate the optimal use of instruments and facilities by enabling the efficient acquisition of high-value datasets. Gaussian process regression (GPR) is a powerful, non-parametric and robust technique for uncertainty quantification and function approximation that can be applied to optimal and autonomous data acquisition. This Review introduces the basics of GPR and discusses several use cases from different fields.

The Pap test is a preventive approach that requires specialized and labor-intensive examination of cytological preparations to track potentially cancerous cells from the internal and external cervix surface. A cytopathologist must analyze many microscopic fields while screening for abnormal cells. Therefore there is hope that a support decision system could assist with clinical diagnosis, for example, by identifying sub-cellular abnormalities, such as changes in the nuclei features. This work proposes an ensemble method for cervical nuclei detection aiming to reduce the workload of cytopathologists. First, a preprocessing phase divides the original image into superpixels, which are input to feature extraction and selection algorithms. The proposed ensemble method combines three classifiers: Decision Tree (DT), Nearest Centroid (NC), and k-Nearest Neighbors (k-NN), which are evaluated against the ISBI’14 Overlapping Cervical Cytology Image Segmentation Challenge dataset. Experiments show that the proposed method is the state-of-the-art algorithm of the literature for recall (0.999) and F1 values (0.993). It produced a recall very close to the optimum value and also kept high precision (0.988).

This paper presents a novel hierarchical nuclei segmentation algorithm for isolated and overlapping cervical cells based on a narrow band level set implementation. Our method applies a new multiscale analysis algorithm to estimate the number of clusters in each image region containing cells, which turns into the input to a narrow band level set algorithm. We assess the nuclei segmentation results on three public cervical cell image databases. Overall, our segmentation method outperformed six state-of-the-art methods concerning the number of correctly segmented nuclei and the Dice coefficient reached values equal to or higher than 0.90. We also carried out classification experiments using features extracted from our segmentation results and the proposed pipeline achieved the highest average accuracy values equal to 0.89 and 0.77 for two-class and three-class problems, respectively. These results demonstrated the suitability of the proposed segmentation algorithm to integrate decision support systems for cervical cell screening.

Fiber-reinforced ceramic-matrix composites are advanced materials resistant to high temperatures, with application to aerospace engineering. Their analysis depends on the detection of embedded fibers, with semi-supervised techniques usually employed to separate fibers within the fiber beds. Here we present an open computational pipeline to detect fibers in ex-situ X-ray computed tomography fiber beds. To separate the fibers in these samples, we tested four different architectures of fully convolutional neural networks. When comparing our neural network approach to a semi-supervised one, we obtained Dice and Matthews coefficients greater than $92.28 \pm 9.65\%$, reaching up to $98.42 \pm 0.03 \%$, showing that the network results are close to the human-supervised ones in these fiber beds, in some cases separating fibers that human-curated algorithms could not find. The software we generated in this project is open source, released under a permissive license, and can be freely adapted and re-used in other domains. All data and instructions on how to download and use it are also available.

The focus of this work is on the detection of nuclei in synthetic images of cervical cells. Finding nuclei is an important step in building a computational method to help cytopathologists identify cell changes from Pap smears. The method developed in this work combines both the Multi-Start and the Iterated Local Search metaheuristics and uses the features of a region to identify a nucleus. It aims to improve the assertiveness of the screening and reduce the professional workload. The irace package was used to automatically calibrate all parameter values of the method. The proposed approach was compared with other methods in the literature according to recall, precision, and F1 metrics using the ISBI Overlapping Cytology Image Segmentation Challenge database (2014). The results show that the proposed method has the second-best values of F1 and recall, while the accuracy is still high.

This paper introduces computational tools for cell classification into normal and abnormal, as well as content-based-image-retrieval (CBIR) for cell recommendation. It also proposes the radial feature descriptors (RFD), which define evenly interspaced segments around the nucleus, and proportional to the convexity of the nuclear boundary. Experiments consider Herlev and CRIC image databases as input to classification via Random Forest and bootstrap: we compare 14 different feature sets by means of False Negative Rate (FNR) and Kappa (kappa), obtaining FNR = 0.02 and kappa = 0.89 for Herlev, and FNR = 0.14 and kappa = 0.78 for CRIC. Next, we sort and rank cell images using convolutional neural networks and evaluate performance with the Mean Average Precision (MAP), achieving MAP = 0.84 and MAP = 0.82 for Herlev and CRIC, respectively. Cell classification show encouraging results regarding RFD, including its sensitivity to intensity variation around the nuclear membrane as it bypasses cytoplasm segmentation. (C) 2019 Elsevier Inc. All rights reserved.

Ceramic matrix composites are resistant materials that withstand high temperatures, but quality control of such composites depends on microtomography image analysis to enable the spatial analysis of fibers, matrix cracks detection, among others. While there are several approaches for fiber detection from microtomography, materials scientists lack computational schemes to validate the accuracy of different fiber detection models. This paper proposes a set of statistical methods to analyse images of CMC in 3D and visualize respective fiber beds, including a lossless data reduction algorithm. The main contribution is our method based on a convolutional neural network that enables evaluation of results from automated fiber detection models, particularly when compared with human curated datasets. We build all the algorithms using free tools to allow full reproducibility of the experiments, and illustrate our results using algorithms designed to probe sample content from gigabyte-size image volumes with minimalistic computational infrastructure.

Incorporating computational fluid dynamics in the design process of jets, spacecraft, or gas turbine engines is often challenged by the required computational resources and simulation time, which depend on the chosen physics-based computational models and grid resolutions. An ongoing problem in the field is how to simulate these systems faster but with sufficient accuracy. While many approaches involve simplified models of the underlying physics, others are model-free and make predictions based only on existing simulation data. We present a novel model-free approach in which we reformulate the simulation problem to effectively increase the size of constrained pre-computed datasets and introduce a novel neural network architecture (called a cluster network) with an inductive bias well-suited to highly nonlinear computational fluid dynamics solutions. Compared to the state-of-the-art in model-based approximations, we show that our approach is nearly as accurate, an order of magnitude faster, and easier to apply. Furthermore, we show that our method outperforms other model-free approaches.

In this work, we propose an Iterated Local Search (ILS) approach to detect cervical cell nuclei from digitized Pap smear slides. The problem consists in finding the best values for the parameters to identify where the cell nuclei are located in the image. This is an important step in building a computational tool to help pathologists to identify cell alterations from Pap tests. Our approach is evaluated by using the ISBI Overlapping Cervical Cytology Image Segmentation Challenge (2014) database, which has 945 synthetic images and their respective ground truth. The precision achieved by the proposed heuristic approach is among the best ones in the literature; however, the recall still needs improvement.

Because of the speed and data rates of time-resolved experiments at facilities such as synchrotron beamlines, automation is critical during time-resolved experiments. In 3D imaging experiments like microCT (CT), this includes recognizing features of interest and “zooming in” spatially and temporally to those features; ideally without requiring advanced information about which features are being imaged. Digital Volume Correlation (DVC) can achieve this by measuring the deformation field between images, but has not been used during autonomous experiments because of the scalability of the codes. In this work, we propose a model for global DVC and a parallel algorithm for solving it for large-scale images, suitable for giving feedback for autonomous experiments at synchrotron-based microCT beamlines. In particular, we leverage recent advancements in entropy-regularized optimal transport to develop efficient, simple-to-implement, parallel algorithms which scale linearly (O(N)) in space and time, where N is the number of voxels, and well with an increasing number of processors. As a demonstration, we compute the deformation field for every voxel from a CT volume with dimensions 2560x2560x2160. We discuss implementation details, drawbacks and future directions.

This work addresses segmentation of volumetric images of woven carbon fiber textiles from micro‐tomography data. We propose a semi‐supervised algorithm to classify carbon fibers that requires sparse input as opposed to completely labeled images. The main contributions are: (a) design of effective discriminative classifiers, for three‐dimensional textile samples, trained on wavelet features for segmentation; (b) coupling of previous step with nonlocal means as simple, efficient alternative to the Potts model; and (c) demonstration of reuse of classifier to diverse samples containing similar content. We evaluate our work by curating test sets of voxels in the absence of a complete ground truth mask. The algorithm obtains an average 0.95 F1 score on test sets and average F1 score of 0.93 on new samples. We conclude with discussion of failure cases and propose future directions toward analysis of spatiotemporal high‐resolution micro‐tomography images.

In this work, we propose a workflow for face recognition under occlusion using the entropy projection from the curved Gabor filter, and create a representative and compact features vector that describes a face. Despite the reduced vector obtained by the entropy projection, it still presents opportunity for further dimensionality reduction. Therefore, we use a Random Forest classifier as an attribute selector, providing a 97% reduction of the original vector while keeping suitable accuracy. A set of experiments using three public image databases: AR Face, Extended Yale B with occlusion and FERET illustrates the proposed methodology, evaluated using the SVM classifier. The results obtained in the experiments show promising results when compared to the available approaches in the literature, obtaining 98.05% accuracy for the complete AR Face, 97.26% for FERET and 81.66% with Yale with 50% occlusion.