Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

General General

X-ModalNet: A semi-supervised deep cross-modal network for classification of remote sensing data.

In ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing (ISPRS)

This paper addresses the problem of semi-supervised transfer learning with limited cross-modality data in remote sensing. A large amount of multi-modal earth observation images, such as multispectral imagery (MSI) or synthetic aperture radar (SAR) data, are openly available on a global scale, enabling parsing global urban scenes through remote sensing imagery. However, their ability in identifying materials (pixel-wise classification) remains limited, due to the noisy collection environment and poor discriminative information as well as limited number of well-annotated training images. To this end, we propose a novel cross-modal deep-learning framework, called X-ModalNet, with three well-designed modules: self-adversarial module, interactive learning module, and label propagation module, by learning to transfer more discriminative information from a small-scale hyperspectral image (HSI) into the classification task using a large-scale MSI or SAR data. Significantly, X-ModalNet generalizes well, owing to propagating labels on an updatable graph constructed by high-level features on the top of the network, yielding semi-supervised cross-modality learning. We evaluate X-ModalNet on two multi-modal remote sensing datasets (HSI-MSI and HSI-SAR) and achieve a significant improvement in comparison with several state-of-the-art methods.

Hong Danfeng, Yokoya Naoto, Xia Gui-Song, Chanussot Jocelyn, Zhu Xiao Xiang

2020-Sep

Adversarial, Cross-modality, Deep learning, Deep neural network, Fusion, Hyperspectral, Label propagation, Multispectral, Mutual learning, Remote sensing, Semi-supervised, Synthetic aperture radar

General General

An automated data extraction and classification pipeline to identify a novel type of neuron within the dorsal striatum based on single-cell patch clamp and confocal imaging data.

In Data in brief

We employed electrophysiological and fluorescence imaging techniques to describe the characteristics of a novel type of neuron discovered in the mouse dorsal striatum. Transgenic mice that express YFP-tagged channelrhodopsin-2 (ChR2) in neurons driven by the promoter for tyrosine hydroxylase (TH) were used and the intrinsic electrical properties of YFP-positive neurons in the dorsal striatum of these mice were characterized using whole-cell patch clamping in acute brain slices. Passive membrane properties - such as membrane capacitance, resting membrane potential and input resistance -and action potential properties- such as amplitude, kinetics and adaptation - were extracted from raw data files. Filling these neurons with neurobiotin enabled visualization of neuronal morphology via immunohistochemical labeling with streptavidin-conjugated fluorophore. Subsequent two-photon imaging allowed analyses of morphological properties such as somaticsize, dendritic branching (Sholl analysis) and density of dendritic spines. Unbiased analyses and hierarchical clustering of both morphological and functional data allowed us to identify a previously undescribed type of striatal neuron with unique properties. To facilitate identification of this new cell type, an end-to-end automated electrophysiology pipeline was developed that extracts relevant parameters and determines striatal neuron identity using neural-network based classifiers. These data and the software tool will permit other investigators to identify this novel type of neuron in their studiesand thereby better understand theroles thatthese neuronsplay in dorsal striatum circuitry.

Mao Miaomiao, Nair Aditya, Augustine George J

2020-Oct

Intrinsic electrical properties, Machine-learning aided classification, Neuron classification, Two-photon imaging, Unsupervised hierarchical clustering, Whole-cell patch clamping

General General

Predicting the Time Period of Extension of Lockdown due to Increase in Rate of COVID-19 Cases in India using Machine Learning.

In Materials today. Proceedings

The research paper proposes a methodology to predict the extension of lockdown in order to eradicate COVID-19 from India. All the concepts related to Coronavirus, its history, prevention and cure is explained in the research paper. Concept used to predict the number of active cases, deaths and recovery is Linear Regression which is an application of machine learning. Extension of lockdown is predicted on the basis of predicted number of active cases, deaths and recovery all over India. To predict the number of active cases, deaths and recovery, date wise analysis of current data was done and necessary parameters like daily recovery, daily deaths, increase rate of covid-19 cases were included. Graphical representation of each analysis and prediction was done in order to make predicted results more understandable. The combined analysis was performed at the end which included the final result of total cases of coronavirus in India. Combined analysis included the no. of cases from start of COVID-19 to the predicted end of cases all over India. [copyright information to be updated in production process].

Wadhwa Parth, Aishwarya Tripathi, Amrendra Singh, Prabhishek Diwakar, Manoj Kumar

2020-Aug-28

COVID-19, COVID-19 pandemic, Coronavirus India, Coronavirus pandemic, coronavirus, lockdown prediction

Dermatology Dermatology

Conceptualising Artificial Intelligence as a Digital Healthcare Innovation: An Introductory Review.

In Medical devices (Auckland, N.Z.)

Artificial intelligence (AI) is widely recognised as a transformative innovation and is already proving capable of outperforming human clinicians in the diagnosis of specific medical conditions, especially in image analysis within dermatology and radiology. These abilities are enhanced by the capacity of AI systems to learn from patient records, genomic information and real-time patient data. Uses of AI range from integrating with robotics to creating training material for clinicians. Whilst AI research is mounting, less attention has been paid to the practical implications on healthcare services and potential barriers to implementation. AI is recognised as a "Software as a Medical Device (SaMD)" and is increasingly becoming a topic of interest for regulators. Unless the introduction of AI is carefully considered and gradual, there are risks of automation bias, overdependence and long-term staffing problems. This is in addition to already well-documented generic risks associated with AI, such as data privacy, algorithmic biases and corrigibility. AI is able to potentiate innovations which preceded it, using Internet of Things, digitisation of patient records and genetic data as data sources. These synergies are important in both realising the potential of AI and utilising the potential of the data. As machine learning systems begin to cross-examine an array of databases, we must ensure that clinicians retain autonomy over the diagnostic process and understand the algorithmic processes generating diagnoses. This review uses established management literature to explore artificial intelligence as a digital healthcare innovation and highlight potential risks and opportunities.

Arora Anmol

2020

artificial intelligence, data, diagnostic algorithms, innovation, machine learning

General General

Comparison of Support Vector Machine, Naïve Bayes and Logistic Regression for Assessing the Necessity for Coronary Angiography.

In International journal of environmental research and public health ; h5-index 73.0

(1) Background: Coronary angiography is considered to be the most reliable method for the diagnosis of cardiovascular disease. However, angiography is an invasive procedure that carries a risk of complications; hence, it would be preferable for an appropriate method to be applied to determine the necessity for angiography. The objective of this study was to compare support vector machine, naïve Bayes and logistic regressions to determine the diagnostic factors that can predict the need for coronary angiography. These models are machine learning algorithms. Machine learning is considered to be a branch of artificial intelligence. Its aims are to design and develop algorithms that allow computers to improve their performance on data analysis and decision making. The process involves the analysis of past experiences to find practical and helpful regularities and patterns, which may also be overlooked by a human. (2) Materials and Methods: This cross-sectional study was performed on 1187 candidates for angiography referred to Ghaem Hospital, Mashhad, Iran from 2011 to 2012. A logistic regression, naive Bayes and support vector machine were applied to determine whether they could predict the results of angiography. Afterwards, the sensitivity, specificity, positive and negative predictive values, AUC (area under the curve) and accuracy of all three models were computed in order to compare them. All analyses were performed using R 3.4.3 software (R Core Team; Auckland, New Zealand) with the help of other software packages including receiver operating characteristic (ROC), caret, e1071 and rminer. (3) Results: The area under the curve for logistic regression, naïve Bayes and support vector machine were similar-0.76, 0.74 and 0.75, respectively. Thus, in terms of the model parsimony and simplicity of application, the naïve Bayes model with three variables had the best performance in comparison with the logistic regression model with seven variables and support vector machine with six variables. (4) Conclusions: Gender, age and fasting blood glucose (FBG) were found to be the most important factors to predict the result of coronary angiography. The naïve Bayes model performed well using these three variables alone, and they are considered important variables for the other two models as well. According to an acceptable prediction of the models, they can be used as pragmatic, cost-effective and valuable methods that support physicians in decision making.

Golpour Parastoo, Ghayour-Mobarhan Majid, Saki Azadeh, Esmaily Habibollah, Taghipour Ali, Tajfard Mohammad, Ghazizadeh Hamideh, Moohebati Mohsen, Ferns Gordon A

2020-Sep-04

angiography, logistic regression, naïve Bayes, support vector machine

Surgery Surgery

Enhancing India's Health Care during COVID Era: Role of Artificial Intelligence and Algorithms.

In Indian journal of otolaryngology and head and neck surgery : official publication of the Association of Otolaryngologists of India

Computerization of health care is the only model to sustain safe health care in this COVID era particularly in overpopulated nations with limited health care providers/systems like India. Accordingly incorporation of computer-based algorithms and artificial intelligence seems very robust and practical models to assist the physician. The advantages of Computerized algorithms to facilitate better screening, diagnosis or follow-up and use of Artificial Intelligence (AI) to aid in medical diagnosis are discussed.

Katyayan Angira, Katyayan Adri, Mishra Anupam

2020-Sep-01

Artificial intelligence, COVID-19, Computer algorithms