Optimized Residual Attention U-Net-Based Lung Cancer Detection

Lung cancer detection using CT pictures is critical for early detection. Deep learning methods can learn hierarchical representations of image characteristics, but detecting subtle patterns and anomalies is difficult. This study proposed an optimized residual attention U-Net for lung tumor detection...

Full description

Saved in:
Bibliographic Details
Published inProceedings (International Confernce on Computational Intelligence and Communication Networks) pp. 984 - 991
Main Authors Kumari, Jyoti, Sinha, Sapna, Singh, Laxman
Format Conference Proceeding
LanguageEnglish
Published IEEE 22.12.2024
Subjects
Accuracy
BGOA (Binary Grasshopper Optimisation Algorithm)
Biomedical imaging
Classification algorithms
Computational modeling
Computed tomography
CT (Computed Tomography) Lung cancer (LC)
Feature extraction
Filtering algorithms
Gaussian Filter
Layers
Lung cancer
Lungs
Optimization
Residual Layers
U-Net
Online AccessGet full text
ISSN2472-7555
DOI10.1109/CICN63059.2024.10847385

Cover

More Information
Summary:Lung cancer detection using CT pictures is critical for early detection. Deep learning methods can learn hierarchical representations of image characteristics, but detecting subtle patterns and anomalies is difficult. This study proposed an optimized residual attention U-Net for lung tumor detection using a Gaussian filter and the Binary Grasshopper Optimisation Algorithm (BGOA). The U-Net enhances feature extraction using multi-attention-gated modules and residual layers, broadening the receptive field while avoiding overfitting. The suggested Optimised Residual Attention U-Net increases classification accuracy by 96.66%, the F1-score by 87.98%, precision by 83.53%, AUC by 98.65%, specificity by 97.22%, and recall by 92.94%. This technique overcomes the issues of detecting subtle patterns and anomalies in CT images and improves the outcomes of lung cancer identification.
ISSN:2472-7555
DOI:10.1109/CICN63059.2024.10847385