Content-based image retrieval for medical infrared images

Area of Science:

• Medical imaging
• Computer vision
• Biomedical engineering

Background:

• Automated processing of medical infrared images has historically focused on specific applications such as breast cancer detection.
• Content-based image retrieval (CBIR) offers a novel approach for analyzing medical thermal images.
• Existing methods often require specialized applications, limiting broader utility.

Purpose of the Study:

• To propose and evaluate the application of content-based image retrieval (CBIR) for medical thermal images.
• To enable retrieval of visually similar thermal images based on extracted image features.
• To facilitate the identification of potential disease similarities through image analysis.

Main Methods:

• Utilizing content-based image retrieval (CBIR) techniques for medical thermal imaging.
• Extracting image features directly from grayscale thermal image data.
• Employing a set of moment invariants as the primary image features for retrieval.

Main Results:

• Demonstrated the feasibility of using CBIR for medical thermal image analysis.
• Showcased that image similarity in CBIR correlates with medical similarities.
• Identified moment invariants as effective features for retrieving relevant thermal images.

Conclusions:

• CBIR is a promising tool for automated analysis of medical thermal images.
• The proposed method using moment invariants can aid in identifying visually and medically similar cases.
• This approach broadens the application of image retrieval in medical thermography.