Multi-View Soft Attention-Based Model for the Classification of Lung Cancer-Associated Disabilities.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Early lung nodule detection improves survival rates but relies on manual, time-consuming radiologist efforts.
• Advanced lung cancer can lead to severe disability, highlighting the need for efficient diagnostic tools.

Purpose of the Study:

• To introduce a Multi-View Soft Attention-Based Convolutional Neural Network (MVSA-CNN) for automated lung nodule classification.
• To classify lung nodules into three categories: benign, primary, and metastatic.

Main Methods:

• Nodule patches were extracted into three distinct views for analysis.
• The MVSA-CNN model was trained and tested using the Lung Image Database Consortium Image Database Resource Initiative (LIDC-IDRI) dataset.
• Model performance was evaluated using a 10-fold cross-validation approach.

Main Results:

• The MVSA-CNN model achieved high performance metrics.
• Achieved 97.10% accuracy, 96.31% sensitivity, and 97.45% specificity.
• Outperformed existing competing methods in lung nodule classification.

Conclusions:

• The MVSA-CNN demonstrates highly predictive performance for lung nodule classification from CT scans.
• This AI model can support more reliable diagnoses, potentially improving patient outcomes.
• The study aims to assist individuals with disabilities who may face healthcare disparities.