Conversion of Pixel to Millimeter in Ultrasound Images: A Methodological Approach and Dataset

Ultrasound imaging is integral to prenatal care, facilitating the identification of fetal landmarks. However, the accuracy of fetal structure measurements largely hinges on the precise conversion of pixel values to millimeters (mm), an essential step in developing artificial intelligence (AI) applications capable of automating fetal measurements. This paper presents a novel pixel intensity-based technique for converting pixel values to millimeters in fetal ultrasound images. Utilizing a publicly available dataset, we have enhanced its value by adding new labels that provide pixel size in millimeters, thereby creating an augmented resource. Our augmented dataset, comprising 2835 fetal head images, represents the largest and most diverse collection of its kind in the literature. The mean pixel size label is 0.144 mm, ranging from a minimum of 0.06 mm to a maximum of 0.33 mm. The proposed technique proves to be accurate and efficient, thus serving as a substantial resource for future AI-based fetal measurement applications in clinical settings.