Convolution Neural Network for Pain Intensity Assessment from Facial Expression.

Area of Science:

• Medical Informatics
• Computer Vision
• Artificial Intelligence

Background:

• Automated pain intensity detection from facial expressions is crucial for smart healthcare due to the subjective nature of pain measurement.
• Existing Convolutional Neural Networks (CNNs) show promise but have limitations in extracting detailed facial features for multi-class pain intensity levels.

Purpose of the Study:

• To develop a highly accurate Deep CNN (DCNN) model for estimating pain intensity from facial expressions.
• To leverage transfer learning for enhanced feature extraction and pain level classification.

Main Methods:

• A DCNN model was proposed, utilizing transfer learning by adapting a pre-trained DCNN and fine-tuning it with facial expression images.
• Experiments were conducted on the UNBC-McMaster shoulder pain archive database.
• Pain intensity was estimated across seven-level thresholds based on facial expressions.

Main Results:

• The proposed DCNN model demonstrated promising improvements in accuracy and performance for pain intensity estimation.
• The method outperformed existing state-of-the-art models in classifying pain intensity levels from facial images.

Conclusions:

• The developed DCNN model with transfer learning offers a more accurate and efficient approach to automated pain intensity detection.
• This technology holds significant potential for improving patient monitoring and healthcare applications.