Figure - available from: PLOS One
This content is subject to copyright.
Chest X-ray images of healthy and infected COVID-19 lungs (a-d) Healthy chest X-ray images. (e-h) COVID-19 chest X-ray images. The images are taken from the COVID-19 Radiography Database [37].
Source publication
In this work, we present a network-based technique for chest X-ray image classification to help the diagnosis and prognosis of patients with COVID-19. From visual inspection, we perceive that healthy and COVID-19 chest radiographic images present different levels of geometric complexity. Therefore, we apply fractal dimension and quadtree as feature...
Citations
... The COVID-19 Radiography Database is particularly comprehensive, including images for COVID-19, normal, and viralpneumonia cases [46]. Its regular updates with new X-ray images ensure that it remains relevant and beneficial for ongoingresearch [47]. ...
Recent advancements in deep learning (DL) have shown significant promise in enhancing diagnostic accuracy (ACC) in medical imaging. This study explores the applicationof Convolutional Neural Networks (CNN) and the MobileNet architecture, optimized with Particle Swarm Optimization (PSO), for the classification of chest X-ray images. Our findings reveal that the CNN achieved impressive classification metrics, with a precision (PER) of 0.94, recall (REC) of 1.00, and an F1-score (F1-s) of 0.97 for the control class. Similarly, for the COVID-19 class, the CNN exhibited a PER of 0.86 and a REC of 0.92, culminating in an F1-s of 0.89. The MobileNet model, prior to PSO optimization, showed remarkable PER and REC across all classes, with overall ACC reaching 0.95. Post-PSO, MobileNet retained an overall ACCof 0.95, with marginal adjustments in PER and REC values, indicating refined model performance. Notably, the control class'sPER improved to 0.99 after PSO, and the COVID-19 class sawan increase in REC to 0.98. These results underscore the potential of using sophisticated Machine Learning (ML) models to aid in the rapid and accurate diagnosis of pulmonary diseases. The high ACC and F1-ss suggest that both CNN and MobileNet models, particularly when enhanced by PSO, could serve as reliable tools in clinical settings, augmenting the capabilities of medical professionals in the interpretation of chest X-rays.
... The COVID-19 Radiography Database is particularly comprehensive, including images for COVID-19, normal, and viralpneumonia cases [46]. Its regular updates with new X-ray images ensure that it remains relevant and beneficial for ongoingresearch [47]. ...
Recent advancements in deep learning (DL) have shown significant promise in enhancing
diagnostic accuracy (ACC) in medical imaging. This study explores the applicationof
Convolutional Neural Networks (CNN) and the MobileNet architecture, optimized with
Particle Swarm Optimization (PSO), for the classification of chest X-ray images. Our findings
reveal that the CNN achieved impressive classification metrics, with a precision (PER) of
0.94, recall (REC) of 1.00, and an F1-score (F1-s) of 0.97 for the control class. Similarly, for
the COVID-19 class, the CNN exhibited a PER of 0.86 and a REC of 0.92, culminating in an
F1-s of 0.89.
The MobileNet model, prior to PSO optimization, showed remarkable PER and REC across
all classes, with overall ACC reaching 0.95. Post-PSO, MobileNet retained an overall ACCof
0.95, with marginal adjustments in PER and REC values, indicating refined model
performance. Notably, the control class’sPER improved to 0.99 after PSO, and the COVID-19
class sawan increase in REC to 0.98.
These results underscore the potential of using sophisticated Machine Learning (ML) models
to aid in the rapid and accurate diagnosis of pulmonary diseases. The high ACC and F1-ss
suggest that both CNN and MobileNet models, particularly when enhanced by PSO, could
serve as reliable tools in clinical settings, augmenting the capabilities of medical professionals
in the interpretation of chest X-rays.
