A Framework to Predict Gastric Cancer Based on Tongue Features and Deep Learning

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Gastric cancer poses a significant global health challenge.
• Early detection and treatment are crucial for patient outcomes.
• Current diagnostic methods for gastric cancer can be invasive and time-consuming.

Purpose of the Study:

• To develop a non-invasive and convenient framework for predicting gastric cancer.
• To explore the relationship between tongue features and gastric cancer.
• To improve early diagnosis rates for gastric cancer.

Main Methods:

• Acquired 703 tongue images using a specialized instrument.
• Applied the Deeplabv3+ network for tongue segmentation.
• Extracted nine tongue features and analyzed them using statistical methods and deep learning.

Main Results:

• Developed a gastric cancer prediction framework combining statistical and deep learning approaches.
• The framework demonstrated strong detection ability with 93.6% accuracy.
• Established a scheme for exploring links between gastric cancer and tongue features.

Conclusions:

• The proposed framework offers an effective method for the early diagnosis of gastric cancer.
• Non-invasive tongue image analysis presents a promising avenue for gastric cancer screening.
• Combining statistical analysis with deep learning enhances predictive accuracy for gastric cancer.