An Improvised Deep-Learning-Based Mask R-CNN Model for Laryngeal Cancer Detection Using CT Images.

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Laryngeal cancer incidence is rising globally, posing diagnostic challenges, especially in advanced stages.
• Current diagnostic tools for laryngeal cancer suffer from limitations like low accuracy in early detection, high computational cost, and time inefficiency.
• Head and neck malignancies, including laryngeal cancer, require precise and timely identification for effective treatment.

Purpose of the Study:

• To introduce a novel, enhanced deep learning model for accurate and efficient laryngeal cancer detection.
• To address the limitations of existing diagnostic methods by improving accuracy and reducing screening time.
• To enable the early identification of minor laryngeal malignancies using real-time image and CT scan analysis.

Main Methods:

• Development and implementation of a deep learning-based Mask R-CNN model.
• Utilization of diverse image datasets and computed tomography (CT) images for real-time analysis.
• Training and validation of the model on the ImageNet dataset.

Main Results:

• The proposed Mask R-CNN model achieved high performance metrics: 98.99% accuracy, 98.99% precision, 97.99% F1 score, and 96.79% recall.
• The model demonstrated capability in detecting minor laryngeal malignancies swiftly and effectively.
• Significant reduction in patient screening time was observed, enabling increased daily patient throughput.

Conclusions:

• The enhanced Mask R-CNN model offers a pragmatic and accurate solution for laryngeal cancer identification.
• The model's speed and accuracy support early detection and timely clinical intervention.
• Further research utilizing extensive datasets holds promise for advancing laryngeal cancer detection methodologies.