Deep ensemble model with blockchain technology for lung cancer detection with secured data sharing.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Blockchain Technology

Background:

• Lung cancer diagnosis is often delayed, leading to high mortality rates.
• Existing methods struggle with feature extraction from CT scans and lack secure data sharing.
• Centralized medical data systems pose risks of breaches and tampering.

Purpose of the Study:

• To propose a secure and efficient framework for lung cancer detection.
• To enhance early diagnosis accuracy using advanced deep learning techniques.
• To ensure privacy-preserving sharing of sensitive medical data.

Main Methods:

• Utilized CT scan data from benchmark datasets.
• Implemented blockchain with smart contracts for secure, decentralized data management.
• Developed a Hybrid Convolutional Neural Network with Autoencoder and Long Short-Term Memory (HCNN-ALSTM) for feature extraction.
• Optimized the HCNN-ALSTM model using the Modified Krill Herd Algorithm (MKHA).

Main Results:

• The MKHA-HCNN-ALSTM model achieved 91.64% accuracy, outperforming existing models.
• Achieved high specificity (92.2%), MCC (84.17), FM (92.44), BM Informedness (84.2), and Markedness (84.13).
• Demonstrated effective feature extraction and parameter optimization for improved diagnostic accuracy.

Conclusions:

• The proposed framework offers a secure and effective solution for early lung cancer detection.
• Blockchain ensures privacy-preserving data sharing among healthcare institutions.
• The MKHA-HCNN-ALSTM model shows significant potential for clinical decision-making and patient care.