Combining LSTM and DenseNet for Automatic Annotation and Classification of Chest X-Ray Images

Abstract

The chest X-ray is a simple and economical medical aid for auxiliary diagnosis and therefore has become a routine item for residents' physical examinations. Based on 40 167 images of chest radiographs and corresponding reports, we explore the abnormality classification problem of chest X-rays by taking advantage of deep learning techniques. First of all, since the radiology reports are generally templatized by the aberrant physical regions, we propose an annotation method according to the abnormal part in the images. Second, building on a small number of reports that are manually annotated by professional radiologists, we employ the long short-term memory (LSTM) model to automatically annotate the remaining unlabeled data. The result shows that the precision value reaches 0.88 in accurately annotating images, the recall value reaches 0.85, and the F1-score reaches 0.86. Finally, we classify the abnormality in the chest X-rays by training convolutional neural networks, and the results show that the average AUC value reaches 0.835.INDEX TERMS Annotation, deep neural network, DenseNet, long short term memory.

AIforCOVID: Predicting the clinical outcomes in patients with COVID-19 applying AI to chest-X-rays. An Italian multicentre study

“…Recently, artificial intelligence (AI) has been widely adopted to analyse CXR for several purposes, such as tuberculosis detection ( Liu et al., 2017 ), abnormality classification and image annotation ( Yan et al., 2019 ), pneumonia screening in pediatric and non pediatric patients ( Radiological Society of North America, 2018 ), edema and fibrosis ( Xu et al., 2018 ). Obviously, the challenge of COVID-19 pandemic has boosted the research efforts of AI in medical imaging and, according to the work presented by Greenspan et al.…”

Quantum transfer learning for breast cancer detection

“…Although there is no validated proof of which method works best for a given clinical problem, Morid et al ( 2020 ) suggest that AlexNET is the most commonly CNN model used for brain magnetic resonance images (Lu et al 2019 ; Wang et al 2019 ) and breast X-rays (Lévy and Jain 2016 ; Omonigho et al 2020 ), while DenseNet for lung X-rays (Liu et al 2019 ; Yan et al 2019 ) and shallow CNN models for skin and dental X-rays (Nunnari et al 2020 ; Zhang et al 2019 ). In addition, with smaller datasets, the most frequently applied TL approach is feature extracting, while fine-tuning is more used with larger datasets; additionally, data augmentation (e.g., translation, rotation of images), as a mean to feed more artificially generated samples to the CNN model in exchange of computational stress, is most frequently employed along with fine-tuning TL approach.…”

Interpretable Pneumonia Detection by Combining Deep Learning and Explainable Models With Multisource Data

“…On the other hand, integrating reports with medical images is used to improve the ability of disease diagnosis [21], [22]. Likewise, [23] trained a small-scale image dataset to diagnose diseases. Still, its results compared using only reports and using only images with increased accuracy of 4% and 7%.…”