Comparative Analysis of Radiological and Machine Learning-based Interpretations for Differentiating COVID-19 and Pneumonia

Authors:
Publisher website
Google Scholar
Abstract
Background: Accurate diagnosis of respiratory conditions is paramount, and this is particularly the case for pneumonia - a common but potentially life-threatening illness that affects many millions worldwide. This review focuses on the diagnostic dilemma and testing paradigm in all types of pneumonia i-e bacterial, viral especially COVID-19 associated. Methods: This study will use chest X-ray and CT scans, traditional tools for pneumonia detection via pulmonary image analysis. Given the subjectivity of radiological interpretations, which may heavily depend on observer expertise, objective methods are required. For solving this problem, we present sophisticated deep learning algorithms to improve image analysis true positive rate and reduce false alarm. This paper compares these state-of-the-art machine-learning techniques with traditional radiological methods to show how technology can revolutionize the diagnosis of pneumonia. Results: The COVID-19 pandemic has presented the complication regarding differentiation of COVID-19-associated pneumonia from than other types due to overlapping symptom and radiological features. We want to characterize these fine differences in our study for even more robust diagnostic accuracy and reliability. Conclusion: We investigated and built a new diagnostic landscape for pneumonia where the traditional individual methods seem to be flawed while machine learning predictions provide some strengths as well as weaknesses. It demonstrates how enhancing diagnosis can again be of par importance towards developing more viably doable public health measures towards mitigating, not only pneumonia, but also other respiratory diseases.

This publication has 32 references indexed in Scilit: