Automated processing of social media content for radiologists: applied deep learning to radiological content on twitter during COVID-19 pandemic

Abstract

Purpose The purpose of this study was to develop an automated process to analyze multimedia content on Twitter during the COVID-19 outbreak and classify content for radiological significance using deep learning (DL). Materials and methods Using Twitter search features, all tweets containing keywords from both "radiology" and "COVID-19" were collected for the period January 01, 2020 up to April 24, 2020. The resulting dataset comprised of 8354 tweets. Images were classified as (i) images with text (ii) radiological content (e.g., CT scan snapshots, X-ray images), and (iii) non-medical content like personal images or memes. We trained our deep learning model using Convolutional Neural Networks (CNN) on training dataset of 1040 labeled images drawn from all three classes. We then trained another DL classifier for segmenting images into categories based on human anatomy. All software used is open-source and adapted for this research. The diagnostic performance of the algorithm was assessed by comparing results on a test set of 1885 images. Results Our analysis shows that in COVID-19 related tweets on radiology, nearly 32% had textual images, another 24% had radiological content, and 44% were not of radiological significance. Our results indicated a 92% accuracy in classifying images originally labeled as chest X-ray or chest CT and a nearly 99% accurate classification of images containing medically relevant text. With larger training dataset and algorithmic tweaks, the accuracy can be further improved. Conclusion Applying DL on rich textual images and other metadata in tweets we can process and classify content for radiological significance in real time.

Mask images on Twitter increase during COVID-19 mandates, especially in Republican counties

“…Moreover, social media image data provide an invaluable and unobtrusive way to measure public health behaviors and trends 58 , 59 . To our knowledge, this is among the earliest work to collect COVID-19 Twitter image data and apply deep learning based image classification 60 , and the first to measure public health behaviors related to COVID-19 with such methods. Future research may employ social media images to identify other behaviors or attributes, such as smoking, alcohol or drug use, obesity, and seat-belt compliance.…”

How social are open-access debates: a follow-up study of tweeters' sentiments

“…Different social networks may differ in their characteristics, functions and purposes. As a worldwide popular microblog, with 500 million users (Haustein, 2019), Twitter was found to be preferred for real-time dissemination of content (Lee, 2018), efficacy and efficiency (Dinata, 2014) and providing information and news sources, live updates (Haug et al, 2021;Khurana et al, 2021;Rony et al, 2018), politics sources (Morini, 2015) and customer service (Sethi et al, 2022). Twitter may provide a suitable platform for the OA debates since it attracts users from both within and outside academia (Bornmann, 2014;Mohammadi et al, 2018;Yu et al, 2019), including publishers (Wang et al, 2021), libraries and universities (Haustein, 2019;Westbury, 2020), and academics and students (Knight and Kaye, 2016).…”

Applications of artificial intelligence in COVID-19 pandemic: A comprehensive review

“…Recently, deep learning-based algorithms have been used by various researchers for combating the COVID-19 pandemic, including convolutional neural network (CNN), recurrent neural network (RNN), and long short-term memory (LSTM) for the COVID-19 detection, diagnosis, classification. Screening, drug repurposing, prediction, and forecasting ( Bogu and Snyder, 2021 , Desai et al, 2020 , Ghoshal and Tucker, 2020 , He et al, 2020 , Hu et al, 2020 , Khurana et al, 2021 , Baig et al, 2019 , Pan et al, 2021 , Sarv Ahrabi et al, 2021 , Sedik et al, 2021 , Soni and Roberts, 2021 ).…”