Cnidaria herd optimized fuzzy C-means clustering enabled deep learning model for lung nodule detection

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Pulmonology

Background:

• Lung nodule detection is vital for lung cancer diagnosis and prevention.
• Existing methods face challenges in scalability, robustness, data availability, and false detection rates.
• Accurate identification of pulmonary nodules is difficult due to variations in shape, size, and location.

Purpose of the Study:

• To develop an effective lung nodule detection model overcoming current limitations.
• To improve the accuracy and reliability of pulmonary nodule identification in medical images.
• To introduce the Cnidaria Herd Optimization algorithm-enabled Bi-directional Long Short-Term Memory (CHSTM) model.

Main Methods:

• Proposed the FC2R segmentation model, integrating optimized fuzzy C-means clustering and Resnet-101 deep learning.
• Employed the Cnidaria Herd Optimization (CHO) algorithm, inspired by krill movement and cnidaria time control, to optimize the CHSTM model.
• Utilized statistical and texture descriptors for feature extraction to enhance detection accuracy.

Main Results:

• The FC2R + CHSTM model achieved 98.09% sensitivity, 97.71% accuracy, and 97.03% specificity on the LUNA-16 dataset.
• On the LIDC/IDRI dataset, the model demonstrated 97.59% accuracy, 96.77% sensitivity, and 98.41% specificity with k-fold validation.
• The proposed model outperformed existing techniques in performance comparisons.

Conclusions:

• The FC2R + CHSTM model offers effective lung nodule detection with high accuracy.
• The research successfully addressed limitations of previous lung nodule detection methods.
• The developed model shows significant potential for clinical application in lung cancer screening.