Related Experiment Video
Updated: Jul 15, 2025
07:30
Electromagnetic Navigation Transthoracic Nodule Localization for Minimally Invasive Thoracic Surgery
Published on: May 4, 2022
3.4K
Point-wise spatial network for identifying carcinoma at the upper digestive and respiratory tract
Lei Zhou1, Huaili Jiang1, Guangyao Li1
1Department of Otorhinolaryngology-Head and Neck Surgery, Zhongshan Hospital Affiliated to Fudan University, Xuhui District, 180 Fenglin Road, , Shanghai, 200032, P. R. China.
BMC Medical Imaging
|September 25, 2023
Summary
Artificial intelligence can now detect head and neck cancers using a single deep learning model. This approach integrates oral pharyngeal, nasopharyngeal, and laryngeal carcinoma detection for improved efficacy.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Artificial intelligence (AI) has shown promise in diagnosing and planning treatments for various cancers.
- Existing AI models are often region-specific for detecting oral pharyngeal, nasopharyngeal, or laryngeal carcinoma.
- A comprehensive AI model for these related head and neck regions is currently lacking.
Purpose of the Study:
- To investigate if a common pattern exists in the cancerous appearance of the oral pharyngeal, nasopharyngeal, and laryngeal regions.
- To develop a single, unified AI model capable of recognizing cancerous patterns across these regions.
- To enhance the efficacy of deep learning models for head and neck cancer detection.
Main Methods:
- Utilized a point-wise spatial attention network model.
- Performed semantic segmentation on images of the oral pharyngeal, nasopharyngeal, and laryngeal regions.
- Trained a single AI model to identify common cancerous features.
Main Results:
- Achieved an average mean Intersection over Union (mIoU) of 86.3%.
- Attained an average pixel accuracy of 96.3%.
- Demonstrated the model's effectiveness in recognizing cancerous patterns.
Conclusions:
- Confirmed that the mucosa and tumors in the oral pharyngeal, nasopharyngeal, and laryngeal regions share recognizable common appearances.
- A single AI model can effectively recognize these shared cancerous features.
- A unified deep learning model can be constructed for the effective detection of these head and neck tumors.
