scholarly journals Quantification of the Immune Content in Neuroblastoma: Deep Learning and Topological Data Analysis in Digital Pathology

2021 ◽  
Vol 22 (16) ◽  
pp. 8804
Author(s):  
Nicole Bussola ◽  
Bruno Papa ◽  
Ombretta Melaiu ◽  
Aurora Castellano ◽  
Doriana Fruci ◽  
...  
Keyword(s):  
Neural Network ◽  
Data Analysis ◽  
Data Augmentation ◽  
Digital Pathology ◽  
Topological Data Analysis ◽  
Main Task ◽  
Learning Approaches ◽  
Training Process ◽  
The Neural Network ◽  
Topological Data

We introduce here a novel machine learning (ML) framework to address the issue of the quantitative assessment of the immune content in neuroblastoma (NB) specimens. First, the EUNet, a U-Net with an EfficientNet encoder, is trained to detect lymphocytes on tissue digital slides stained with the CD3 T-cell marker. The training set consists of 3782 images extracted from an original collection of 54 whole slide images (WSIs), manually annotated for a total of 73,751 lymphocytes. Resampling strategies, data augmentation, and transfer learning approaches are adopted to warrant reproducibility and to reduce the risk of overfitting and selection bias. Topological data analysis (TDA) is then used to define activation maps from different layers of the neural network at different stages of the training process, described by persistence diagrams (PD) and Betti curves. TDA is further integrated with the uniform manifold approximation and projection (UMAP) dimensionality reduction and the hierarchical density-based spatial clustering of applications with noise (HDBSCAN) algorithm for clustering, by the deep features, the relevant subgroups and structures, across different levels of the neural network. Finally, the recent TwoNN approach is leveraged to study the variation of the intrinsic dimensionality of the U-Net model. As the main task, the proposed pipeline is employed to evaluate the density of lymphocytes over the whole tissue area of the WSIs. The model achieves good results with mean absolute error 3.1 on test set, showing significant agreement between densities estimated by our EUNet model and by trained pathologists, thus indicating the potentialities of a promising new strategy in the quantification of the immune content in NB specimens. Moreover, the UMAP algorithm unveiled interesting patterns compatible with pathological characteristics, also highlighting novel insights into the dynamics of the intrinsic dataset dimensionality at different stages of the training process. All the experiments were run on the Microsoft Azure cloud platform.

Model discrepancy of Earth polar motion using topological data analysis and convolutional neural network analysis

2020 ◽  
Vol 31 (08) ◽  
pp. 2050117
Author(s):  
Dongjin Lee ◽  
Christopher Bresten ◽  
Kookhyoun Youm ◽  
Ki-Weon Seo ◽  
Jae-Hun Jung
Keyword(s):  
Neural Network ◽  
Data Analysis ◽  
Convolutional Neural Network ◽  
Polar Motion ◽  
Topological Data Analysis ◽  
Accurate Analysis ◽  
Long Baseline ◽  
Motion Signal ◽  
Polar Motion Excitation ◽  
Topological Data

An accurate analysis of the polar motion variation is essential to understand the global change of the environment and predict useful information about short-term and long-term change in climate. Observation of polar motion excitation using multiple measurements including Very-Long-Baseline-Interferometry (VLBI) provides highly accurate measurement of polar motion variation. The observed polar motion excitation has been modeled with multiple geophysical models, but the discrepancies between observations and models still exist. In this paper, we propose two approaches for detecting the discrepancy of the polar motion excitation: topological data analysis (TDA) and convolutional neural network (CNN) analysis. Our methods clearly show that the observed polar motion has a different topological structure from the model data, and there are time periods that the model fails to represent the polar motion. Numerical results indicate that the proposed methods show promise for applications to polar motion signal analysis.

scholarly journals The Topology of Neuronal Structures Exposed to Cosmic Radiation

2020 ◽  
Vol 19 ◽  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Data Analysis ◽  
Cosmic Radiation ◽  
Network Code ◽  
Topological Data Analysis ◽  
Analysis Approach ◽  
The Neural Networks ◽  
Topological Data

In this paper, we focus on some leader NASA experiences to explore how cosmic radiation caused significant reductions in dendrite and spine complexity. We adopt a topological data analysis approach and extract more information then the classical methods. Our key idea is to use the NASA images of the neural networks of some mouses that were exposed 12 weeks to cosmic radiation. We associate to this neural network code bares that give us more information, that that given by the original experiences.

scholarly journals Topological Data Analysis Approaches to Uncovering the Timing of Ring Structure Onset in Filamentous Networks

2021 ◽  
Vol 83 (3) ◽  
Author(s):  
Maria-Veronica Ciocanel ◽  
Riley Juenemann ◽  
Adriana T. Dawes ◽  
Scott A. McKinley
Keyword(s):  
Time Series ◽  
Data Analysis ◽  
Actin Filaments ◽  
Time Series Data ◽  
Polymer Networks ◽  
Topological Data Analysis ◽  
Series Data ◽  
Topological Features ◽  
Topological Data

AbstractIn developmental biology as well as in other biological systems, emerging structure and organization can be captured using time-series data of protein locations. In analyzing this time-dependent data, it is a common challenge not only to determine whether topological features emerge, but also to identify the timing of their formation. For instance, in most cells, actin filaments interact with myosin motor proteins and organize into polymer networks and higher-order structures. Ring channels are examples of such structures that maintain constant diameters over time and play key roles in processes such as cell division, development, and wound healing. Given the limitations in studying interactions of actin with myosin in vivo, we generate time-series data of protein polymer interactions in cells using complex agent-based models. Since the data has a filamentous structure, we propose sampling along the actin filaments and analyzing the topological structure of the resulting point cloud at each time. Building on existing tools from persistent homology, we develop a topological data analysis (TDA) method that assesses effective ring generation in this dynamic data. This method connects topological features through time in a path that corresponds to emergence of organization in the data. In this work, we also propose methods for assessing whether the topological features of interest are significant and thus whether they contribute to the formation of an emerging hole (ring channel) in the simulated protein interactions. In particular, we use the MEDYAN simulation platform to show that this technique can distinguish between the actin cytoskeleton organization resulting from distinct motor protein binding parameters.

scholarly journals Classification of apatite structures via topological data analysis: a framework for a ‘Materials Barcode’ representation of structure maps

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Scott Broderick ◽  
Ruhil Dongol ◽  
Tianmu Zhang ◽  
Krishna Rajan
Keyword(s):  
Machine Learning ◽  
Data Analysis ◽  
Crystal Chemistry ◽  
Persistent Homology ◽  
Hierarchical Classification ◽  
Topological Data Analysis ◽  
Learning Tool ◽  
Coordination Polyhedra ◽  
Machine Learning Tool ◽  
Topological Data

AbstractThis paper introduces the use of topological data analysis (TDA) as an unsupervised machine learning tool to uncover classification criteria in complex inorganic crystal chemistries. Using the apatite chemistry as a template, we track through the use of persistent homology the topological connectivity of input crystal chemistry descriptors on defining similarity between different stoichiometries of apatites. It is shown that TDA automatically identifies a hierarchical classification scheme within apatites based on the commonality of the number of discrete coordination polyhedra that constitute the structural building units common among the compounds. This information is presented in the form of a visualization scheme of a barcode of homology classifications, where the persistence of similarity between compounds is tracked. Unlike traditional perspectives of structure maps, this new “Materials Barcode” schema serves as an automated exploratory machine learning tool that can uncover structural associations from crystal chemistry databases, as well as to achieve a more nuanced insight into what defines similarity among homologous compounds.

scholarly journals Rethinking the Random Cropping Data Augmentation Method Used in the Training of CNN-Based SAR Image Ship Detector

2020 ◽  
Vol 13 (1) ◽  
pp. 34
Author(s):  
Rong Yang ◽  
Robert Wang ◽  
Yunkai Deng ◽  
Xiaoxue Jia ◽  
Heng Zhang
Keyword(s):  
Neural Network ◽  
Data Augmentation ◽  
Back Propagation ◽  
Detection Performance ◽  
Training Data ◽  
Sar Image ◽  
Optical Images ◽  
The Neural Network ◽  
Effective Training ◽  
Standard Configuration

The random cropping data augmentation method is widely used to train convolutional neural network (CNN)-based target detectors to detect targets in optical images (e.g., COCO datasets). It can expand the scale of the dataset dozens of times while consuming only a small amount of calculations when training the neural network detector. In addition, random cropping can also greatly enhance the spatial robustness of the model, because it can make the same target appear in different positions of the sample image. Nowadays, random cropping and random flipping have become the standard configuration for those tasks with limited training data, which makes it natural to introduce them into the training of CNN-based synthetic aperture radar (SAR) image ship detectors. However, in this paper, we show that the introduction of traditional random cropping methods directly in the training of the CNN-based SAR image ship detector may generate a lot of noise in the gradient during back propagation, which hurts the detection performance. In order to eliminate the noise in the training gradient, a simple and effective training method based on feature map mask is proposed. Experiments prove that the proposed method can effectively eliminate the gradient noise introduced by random cropping and significantly improve the detection performance under a variety of evaluation indicators without increasing inference cost.

Sign in / Sign up

Export Citation Format

Share Document