PhotIQA: A photoacoustic image data set with image quality ratings

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Image quality assessment (IQA) is vital for evaluating new imaging algorithms.
• Existing IQA methods, developed for natural images, show limitations when applied to medical images due to differing properties.
• Standardized benchmarking for photoacoustic imaging (PAI) quality is currently lacking.

Purpose of the Study:

• To address the need for robust IQA measures in medical imaging, especially for PAI.
• To develop and validate a comprehensive dataset for training and testing IQA algorithms.
• To facilitate the advancement of image reconstruction and quality evaluation in PAI.

Main Methods:

• Assembled PhotIQA, a dataset comprising 1134 photoacoustic images.
• Collected expert ratings on five distinct quality properties for each image.
• Utilized a full-reference IQA approach for detailed quality assessment.

Main Results:

• The PhotIQA dataset provides a valuable resource for IQA research.
• Expert ratings offer detailed insights into image quality across multiple dimensions.
• The dataset's availability supports the development of PAI-specific IQA metrics.

Conclusions:

• The PhotIQA dataset is crucial for advancing IQA in photoacoustic imaging.
• This resource enables the development of more accurate and reliable IQA measures for medical applications.
• Public availability on Zenodo promotes collaborative research and innovation in medical image analysis.