Related Experiment Video
Updated: Jan 7, 2026
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
2.2K
Swallow Winged Kite Optimization with Shuffle Attention Xtreme Gradient Boost Network for lung cancer detection using
1Department of Computer Science and Engineering, R.M.K. College of Engineering and Technology, R.S.M. Nagar, Puduvoyal, 601206, Tamil Nadu, India. sandhiyacse@rmkcet.ac.in.
Summary
A novel deep learning model, SWKO_SA-XGBNet, enhances early lung cancer detection from CT images. This method significantly improves diagnostic accuracy, offering a more sensitive and less invasive approach for timely intervention.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Lung cancer diagnosis is often delayed due to late-stage detection and lack of early symptoms.
- Current diagnostic methods are invasive, costly, and lack the sensitivity for effective early detection.
- Advanced stage diagnosis limits treatment options and negatively impacts patient outcomes.
Purpose of the Study:
- To develop an advanced deep learning model for accurate and early lung cancer detection using CT images.
- To address the limitations of conventional diagnostic techniques by proposing a more sensitive and non-invasive approach.
- To improve patient outcomes through timely and precise lung cancer diagnosis.
Main Methods:
- A novel Swallow Winged Kite Optimization with Shuffle Attention Xtreme Gradient Boost Network (SWKO_SA-XGBNet) was designed for lung cancer detection.
- Image preprocessing involved the Non-Local Means (NLM) filter, followed by lung nodule segmentation using Medical Images Segment Net (MIS-Net).
- The SWKO_SA-XGBNet model integrates Convolutional Xtreme Gradient Boost (ConvXGB), Fractional Calculus (FC), and Shuffle Attention Network (SA-Net), optimized via a hybrid SWKO algorithm.
Main Results:
- The SWKO_SA-XGBNet model achieved high diagnostic performance on CT images.
- Key performance metrics included an Accuracy of 96.88%, True Positive Rate (TPR) of 97.46%, True Negative Rate (TNR) of 96.60%, Precision of 96.01%, and F1-score of 96.73%.
- The proposed method demonstrated superior sensitivity and specificity in identifying lung nodules.
Conclusions:
- The developed SWKO_SA-XGBNet model offers a highly accurate and efficient approach for early lung cancer detection.
- This AI-driven method shows significant potential to overcome the limitations of traditional diagnostic tools.
- The findings suggest a promising advancement in oncological imaging for improved patient survival rates.
