An Advanced Lung Carcinoma Prediction and Risk Screening Model Using Transfer Learning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Lung cancer (lung carcinoma) poses a significant mortality risk, emphasizing the need for early diagnosis.
• Current lung cancer prediction models struggle with accuracy, noise, and low contrast in medical images.
• Effective early detection and risk assessment are crucial for improving patient outcomes.

Purpose of the Study:

• To propose an advanced lung carcinoma prediction and risk screening model utilizing transfer learning.
• To address limitations of existing models, including low accuracy, noise, and low contrast in computed tomography (CT) images.
• To enhance early lung cancer diagnosis and risk stratification.

Main Methods:

• Preprocessing of lung CT images: noise removal, contrast stretching, convex hull lung region extraction, and edge enhancement.
• Image segmentation using the modified Bates distribution coati optimization (B-RGS) algorithm for feature extraction.
• Classification of lung cancer as normal or abnormal using a PResNet classifier, followed by risk screening (low/high risk).

Main Results:

• The proposed model achieved high performance metrics comparable to state-of-the-art methods.
• Achieved accuracy of 98.21%, precision of 98.71%, and recall of 97.46%.
• Demonstrated effectiveness in early lung carcinoma prediction and risk assessment.

Conclusions:

• The developed transfer learning model offers an efficient and effective solution for early lung cancer detection.
• The methodology shows significant promise in improving the accuracy and reliability of lung carcinoma risk screening.
• This approach validates the potential of advanced AI techniques in enhancing oncological diagnostics.