Automatic Identification of Depression Using Facial Images with Deep Convolutional Neural Network

Area of Science:

• Computer Science
• Medical Imaging
• Psychiatry

Background:

• Depression affects approximately 280 million people globally, necessitating innovative diagnostic tools.
• Distinct facial features associated with depression offer potential for automated recognition.
• Deep convolutional neural networks (CNNs) are emerging as powerful tools for image analysis and classification.

Purpose of the Study:

• To develop and evaluate a CNN-based method for automatic depression recognition using facial images.
• To compare the performance of various CNN architectures in classifying depression from facial data.
• To establish the efficacy of automated facial image analysis for depression screening.

Main Methods:

• A dataset of 1020 depressed patients and 1100 healthy participants' facial images was curated.
• Images were divided into training, testing, and validation sets at a 7:2:1 ratio.
• Several CNN models, including Fully Connected Convolutional Neural Network (FCN), VGG11, VGG19, ResNet50, and Inception v3, were trained and evaluated.

Main Results:

• The FCN model achieved the highest accuracy (98.23%) and precision (98.11%).
• VGG19 and Inception v3 models also demonstrated high performance, with accuracies of 97.35% and 97.10%, respectively.
• All tested CNN models showed significant potential in distinguishing between depressed patients and healthy individuals.

Conclusions:

• Deep convolutional neural networks provide a viable approach for rapid, accurate, and automated depression identification.
• Facial image analysis using CNNs can serve as a valuable supplementary tool in mental health diagnostics.
• Further research can explore larger datasets and diverse populations to enhance model generalizability.