Thermal Evaluation of Therapeutic Effects
This is the GitHub repository for the project "Thermal Evaluation System for Therapeutic Effects".
Description
• Analyzed patients’ facial infrared thermal image with image recognition and processing technology, utilized digital methods to quantify patients’ temperature distribution and indication changes before and after treatment, ensuring accurate diagnoses and treatments and reducing doctor workloads.
• Carried out image pre-processing with filtering-based image denoising and adaptive threshold.
• Employed the Detectron2 framework to segment patients’ faces and filter out irrelevant areas, carried out image registration based on the segmentation results to enable further analysis.
• Applied digital image processing methods to the segmented areas and proposed quantifiable metrics to evaluate temperature changes, eventually integrating these functions into a diagnosis system.
• Wrapped up our work and applied for a Utility Model Patent in China.
How to Run
-
Set up Python environment and install Detectron2.
-
Get the pre-trained model here and put it under
output -
Put target thermal images in
datasets/raw -
Run the project using the command:
python main.py
Key Features
Enabling comparison between images before and after treatment
The system facilitates clinical diagnoses by extracting digital features from the corresponding areas of the images before and after treatment.
Quantifiable metrics for digital evaluation
Including highest/average temperature, temperature difference/gradient, isotherm, etc.
Partition of face area according to clinical study
Divide the face area into twelve regions: left/right forehead, canthuses, cheek, nasal wing, mouth corner.
Mark the points with the highest temperatures (blue asterisk) and the areas that exceed the temperature thresholds (in cyan).
Show isotherms according to customized parameters.
Show difference image.
Show gradience image.
Customizable parameters for quantifiable evaluation
The temperature thresholds and isotherm parameters, along with other configurations, can be adjusted by user through the interface.
Intuitive interface for easy evaluation
Displays the images before and after treatment in a way that allows for direct comparison and analysis of therapeutic effects.