Comparison of some CNN architectures for detecting cardiomegaly from chest X-ray images

In medical image analysis, deep learning and convolutional neural networks (CNN) are widely employed, particularly in tasks such as classification and segmentation.  This study specifically addresses their application in healthcare for detecting cardiomegaly, a condition characterized by an enlarged heart, often related to factors such as hypertension or coronary artery diseases.  The primary objective is to develop an algorithm to identify cardiomegaly in chest X-ray images, constituting a binary classification problem (whether the image exhibits cardiomegaly or not).  Using the CXR8 dataset from the National Institute of Health Clinical Center, comprising 2\,776 cardiomegaly images and 60\,361 no finding images, the inputs are labeled images, and the outputs are the corresponding labels (Cardiomegaly or No Finding).  Employing Keras and TensorFlow Python libraries, we aim to construct a CNN model that excels in binary classification, distinguishing between cardiomegaly and no finding in chest X-ray images.

глибинне навчання
convolutional neural networks
cardiomegaly
бінарна класифікація
  1. Candemir S., Rajaraman S., Thoma G., Antani S.  Deep learning for grading cardiomegaly severity in chest X-rays: An investigation.  2018 IEEE Life Sciences Conference (LSC).  109–113 (2018).
  2. Sogancioglu E., Murphy K., Calli E., Scholten E. T., Schalekamp S., Van Ginneken B.  Cardiomegaly detection on chest radiographs: Segmentation versus classification.  IEEE Access.  8, 94631–94642 (2020).
  3. Zhou S., Zhang X., Zhang R.  Identifying cardiomegaly in ChestX-ray8 using transfer learning.  MEDINFO 2019: Health and Wellbeing e-Networks for All.  264, 482–486 (2019).
  4. Yanar E., Hardalaç F.  Clinical Decision Support for Early Diagnosis of Cardiomegaly by Using Deep Learning Techniques on Chest X-rays.  Computing in Cardiology. 42 (2023).
  5. Hamedian M.  Automatic classification of cardiomegaly using deep convolutional neural network (2022).
  6. Grant D., Papież B. W., Parsons G., Tarassenko L., Mahdi A.  Deep learning classification of cardiomegaly using combined imaging and non-imaging ICU data.  Medical Image Understanding and Analysis: 25th Annual Conference, MIUA 2021, Oxford, United Kingdom, July 12–14, 2021. 547–558 (2021).
  7. Lee M. S., Kim Y. S., Kim M., Usman M., Byon S. S., Kim S. H., Lee B. I., Lee B. D.  Evaluation of the feasibility of explainable computer-aided detection of cardiomegaly on chest radiographs using deep learning.  Scientific Reports.  11 (1), 16885 (2021).
  8. Kim M.-J., Kim J.-H.  Development of convolutional neural network model for classification of cardiomegaly X-ray images.  Journal of Mechanics in Medicine and Biology.  22 (08), 2240020 (2022).
  9. Bar Y., Diamant I., Wolf L., Greenspan H.  Deep learning with non-medical training used for chest pathology identification.  Medical Imaging 2015: Computer-Aided Diagnosis.  9414, 215–221 (2015).
  10. Rubin J., Sanghavi D., Zhao C., Lee K., Qadir A., Xu-Wilson M.  Large scale automated reading of frontal and lateral chest X-rays using dual convolutional neural networks.  Preprint arXiv:1804.07839 (2018).
  11. Chen L., Mao T., Zhang Q.  Identifying cardiomegaly in chest X-rays using dual attention network.  Applied Intelligence.  52 (10), 11058–11067 (2022).
  12. Rajaraman S., Kim I., Antani S. K.  Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles.  PeerJ.  8, e8693 (2020).
  13. Bar Y., Diamant I., Wolf L., Lieberman S., Konen E., Greenspan H.  Chest pathology identification using deep feature selection with non-medical training.  Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization.  6 (3), 259–263 (2018).
  14. Calin O.  Deep Learning Architectures – A Mathematical Approach.  Springer Series in the Data Sciences, Springer (2020).
  15. Wu J. X., Pai C. C., Kan C. D., Chen P. Y., Chen W. L., Lin C. H.  Chest X-ray image analysis with combining 2D and 1D convolutional neural network based classifier for rapid cardiomegaly screening.  IEEE Access.  10, 47824–47836 (2022).
  16. Kim M. J., Kim J. H.  Proposal of a Convolutional Neural Network Model for the Classification of Cardiomegaly in Chest X-ray Images.  Journal of the Korean Society of Radiology.  15 (5), 613–620 (2021).
  17. Yoo H., Han S., Chung K.  Diagnosis support model of cardiomegaly based on CNN using ResNet and explainable feature map.  IEEE Access.  9, 55802–55813 (2021).
  18. Junior J. R. F., Cardenas D. A. C., Moreno R. A., de S{\'a} Rebelo M. d. F., Krieger J. E., Gutierrez M. A.  A general fully automated deep-learning method to detect cardiomegaly in chest X-rays.  Medical Imaging 2021: Computer-Aided Diagnosis.  11597, 537–542 (2021).
  19. Jeong W. Y., Kim J. H.  A Study on the Improvement of Accuracy of Cardiomegaly Classification Based on InceptionV3.  Journal of Biomedical Engineering Research.  43 (1), 45–51 (2022).
  20. Bougias H., Georgiadou E., Malamateniou C., Stogiannos N.  Identifying cardiomegaly in chest X-rays: a cross-sectional study of evaluation and comparison between different transfer learning methods.  Acta Radiologica.  62 (12), 1601–1609 (2021).