EP26.18: 3D visualisation of uterine myomas using Augmented Reality Guided System ultrasound image

Augmented Reality Guided System (ARGS) ultrasound image is an ultrasound imaging with augmented reality (AR) technology with deep learning (DL) based semantic segmentation. It can bring ultrasound images into real-world space, i.e. patient's body space, by providing the image position and pose information. Using the ARGS ultrasound image, physicians can sense where and how the ultrasound image is being acquired. We aimed to evaluate the accuracy of ARGS ultrasound image for uterine myomas. We evaluated 27 patients who had uterine myomas. A deep learning model detects and segments 48 uterine lesions out of the 27 patients. 40 lesions out of 23 patients were used for training, while the rest of the lesions were used for validation and testing. This experiment uses accuracy for how accurately it finds suitable pixels and the Dice score for how accurately it gets the correct shapes of the uterine myomas. Deep learning-based segmentation model could accurately define ultrasound images for uterine myomas lesions. The mean accuracy out of the experiment was higher than 0.96. The Dice score ranged from 0.4203 to 0.9114, which was produced with less training data, and as the training went on, the Dice score gradually improved. ARGS ultrasound image was able to visualise 3D reconstructed images for uterine myomas (figure 1). ARGS ultrasound image visualises the live ultrasound image and segmented myoma surface in 3D while scanning, which can give size, shape, positional and realistic information. ARGS is expected to facilitate the diagnosis and help with treatment plans, including surgery. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.