Multi-Modal Data Fusion and Dual Validation for Lung Lesion Detection and Optimization

Abstract

The research proposes Dual Data Validation (DDV) and Whale Optimization Algorithm (WOA) which combines CT and Ultrasound patterns to boost medical imaging effectiveness and precision during lung lesion detection. CT lung datasets undergo Deep Neural network (DNN) framework processing during the primary stage to extract critical features by means of attribute co-relationship mapping. The WOA optimization of the feature matrix leads to effective feature selection while enabling primary decision-making to produce an optimized trained dataset for dual feature verification. The parallel processing system extracts features from ultrasound samples through MOTIF feature extraction while using Swarm Optimization Terminology for extensive multi-modal validation. The DDV technique starts through threshold mapping techniques applied to obtained datasets to calculate Regions of Interest (ROI) and undertake optimization processes for precise decision-making. The proposed method reaches an 87.32% accurate decision support rate through evaluation on a dataset of 1794 CT and Ultrasound images from Kaggle. The technique achieves a 12.73% better prediction accuracy than regular uni-processing approaches because it combines dual modality data with optimization methods to enhance lung lesion detection.