Lung disease detection utilising HOG features with SCA-ELM model with FRFLICM segmentation in healthcare systems

Lung infections increase the mortality rate in the present day due to environmental pollution. The infection's size, shape, and position differ from a dissimilar patient's lung. It becomes problematic for a clinical physician to detect infected regions of the lungs from X-ray images. It was challenging for the medical practitioner to segment, detect, and extract infected lung areas from X-ray images. This research suggests a histogram oriented graph (HOG) feature-based hybrid modified sine cosine algorithm (SCA)-extreme learning machine model to classify infected and non-infected lung diseases. This technique was proposed by utilising modified sine cosine optimisation to increase the quality of images. The segmentation accuracy proposed improved the quality measures such as SSIM and PSNR were obtained as 0.9878 and 35.26, respectively. The proposed SCA-ELM classifier achieved 98.78%, sensitivity, 99.23% specificity, 99.26% accuracy, and 24.1128 seconds computational time with the proposed SCA-ELM models.