Scholarlink

<p>Chronic exposure to crystalline silica dust at the workplace may lead to silicosis, a severe and irreversible lung condition that is one of the significant occupational risks in the world. With chest radiographs, to help radiologists with silicosis screening and staging, the present research proposes to create an AI-based model. Using artificial intelligence (AI) and related technologies to diagnose diseases proactively has been a fascinating field of study throughout the past ten years. Pre-processing was performed on images from the publicly available dataset Kaggle dataset and radiopaedia.org, the modified Unet method was used to segment the images and the nature-inspired hybrid metaheuristic Black Widow Optimisation with Dragonfly algorithm (BWODA) algorithm was used to select a subset of the most relevant features. Popular classifiers for multiclass classification included the Support Vector Machine (SVM), Random Forest (RF), and Decision Tree (DT). The proposed system’s performance has been evaluated against three additional well-known optimisation algorithms: Black widow optimisation (BWO), Ant Colony Optimisation (ACO), and Particle Swarm Optimisation (PSO). Evaluation metrics for this research include the Accuracy, Precision, Recall, F1 score, Specificity, Kappa value, Jaccard Index and Matthews Correlation Coefficient. Random Forest method outperformed better than other classifiers when fed with the optimised features derived from the proposed hybrid technique BWODA with 95.41%, 94.22%, 94.24%, 97.93%, 95.23%, 97.41%, 93.23% and 94.32%.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Hybrid Meta-Heuristics Approach for the detection and classification of Silicosis

  • 期刊论文

  • 发表日期：

    2026年3月3日

查看全文
  • N. Shivaanivarsha,
  • P. Kavipriya

摘要

Chronic exposure to crystalline silica dust at the workplace may lead to silicosis, a severe and irreversible lung condition that is one of the significant occupational risks in the world. With chest radiographs, to help radiologists with silicosis screening and staging, the present research proposes to create an AI-based model. Using artificial intelligence (AI) and related technologies to diagnose diseases proactively has been a fascinating field of study throughout the past ten years. Pre-processing was performed on images from the publicly available dataset Kaggle dataset and radiopaedia.org, the modified Unet method was used to segment the images and the nature-inspired hybrid metaheuristic Black Widow Optimisation with Dragonfly algorithm (BWODA) algorithm was used to select a subset of the most relevant features. Popular classifiers for multiclass classification included the Support Vector Machine (SVM), Random Forest (RF), and Decision Tree (DT). The proposed system’s performance has been evaluated against three additional well-known optimisation algorithms: Black widow optimisation (BWO), Ant Colony Optimisation (ACO), and Particle Swarm Optimisation (PSO). Evaluation metrics for this research include the Accuracy, Precision, Recall, F1 score, Specificity, Kappa value, Jaccard Index and Matthews Correlation Coefficient. Random Forest method outperformed better than other classifiers when fed with the optimised features derived from the proposed hybrid technique BWODA with 95.41%, 94.22%, 94.24%, 97.93%, 95.23%, 97.41%, 93.23% and 94.32%.