Classification of chest radiographs into healthy/pneumonia using Harris-Hawks Algorithm optimized deep-features

Abstract Pneumonia is a pulmonary infection that causes thoracic discomfort, typically caused by bacteria, or viruses. The pneumonia in children and elderly is medical emergency and hence appropriate diagnosis and treatment is necessary. Clinical-level screening of pneumonia is frequently executed u...

Full description

Saved in:
Bibliographic Details
Main Authors: K. Vijayakumar, Mohammad Nazmul Hasan Maziz, Swaetha Ramadasan, Seifedine Kadry, S. Arunmozhi
Format: Article
Language:English
Published: Springer 2025-06-01
Series:Discover Computing
Subjects:
Online Access:https://doi.org/10.1007/s10791-025-09629-8
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Pneumonia is a pulmonary infection that causes thoracic discomfort, typically caused by bacteria, or viruses. The pneumonia in children and elderly is medical emergency and hence appropriate diagnosis and treatment is necessary. Clinical-level screening of pneumonia is frequently executed using the chest X-ray and its analysis will help in treatment planning and execution. Recently, several pre-trained deep-learning (PDL) based systems are developed to identify disease in different imaging modalities, including the chest X-ray. This study aims to develop a PDL-based tool to analyse chest X-ray dataset to identify the pneumonia. This PDL-tool performs the following tasks on the X-ray database; (i) detection of healthy/pneumonia, and (ii) detecting the viral/bacterial pneumonia. Along with the traditional deep-features based classification using the SoftMax, this work also considered Harris-Hawks Algorithm (HHA) algorithm based features optimization and serial features integration to generate fused-features vector (FFV). The experimental outcome authenticates that this PDL-tool helps to offer improved accuracy with the HHA-optimized features. This work provided an accuracy of 99.3750% during healthy/pneumonia detection with FFV and Support Vector Machine (SVM), and detection accuracy of 88.5417% during viral/bacterial pneumonia detection with FFV and SVM.
ISSN:2948-2992