TSSG-CNN: A Tuberculosis Semantic Segmentation-Guided Model for Detecting and Diagnosis Using the Adaptive...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Science

Background:

• Tuberculosis (TB) is a contagious infectious disease caused by Mycobacterium, primarily affecting the lungs but potentially other organs.
• TB spreads through airborne droplets from infected individuals, necessitating effective diagnostic tools.
• Current diagnostic methods require enhancement for improved accuracy and early detection.

Purpose of the Study:

• To develop and evaluate a novel deep learning model for enhanced tuberculosis detection using chest X-ray images.
• To improve the precision of TB diagnosis by combining semantic segmentation with a convolutional neural network (CNN) architecture.
• To assess the performance of the proposed Tuberculosis Segmentation-Guided Diagnosis Model (TSSG-CNN) against other established models.

Main Methods:

• Proposed a novel Tuberculosis Segmentation-Guided Diagnosis Model (TSSG-CNN) integrating semantic segmentation and adaptive CNN.
• Utilized a combined deep learning segmentation and classification model for precise chest X-ray image analysis.
• Employed the Mayfly Algorithm (MA) for simplified feature optimization and compared TSSG-CNN with simple CNN, BN-CNN, and DCNN.

Main Results:

• The TSSG-CNN model achieved a high accuracy of 98.75% and an F1 score of 98.70%.
• The proposed model significantly outperformed other evaluated models, including simple CNN, BN-CNN, and DCNN.
• Evaluation findings confirmed the effectiveness of the deep learning segmentation model within the TSSG-CNN architecture.

Conclusions:

• The TSSG-CNN model represents a highly accurate strategy for tuberculosis detection.
• The study highlights the potential of the TSSG-CNN model for precise and early diagnosis of TB.
• Further research is warranted to explore the full capabilities of this deep learning approach in medical diagnostics.