Automatic Prostate Volume Estimation in Transabdominal Ultrasound Images

• DOI: 10.48550/arxiv.2502.07859
• Publication Date: 2025-02-11
• Authors: Tiziano Natali; Liza M. Kurucz; Matteo Fusaglia; Laura S. Mertens; Theo J. M. Ruers; Pim J van Leeuwen; Behdad Dashtbozorg
• Keywords: Transabdominal ultrasound

Prostate cancer is a leading health concern among men, requiring accurate and accessible methods for early detection risk stratification. volume (PV) key parameter in multivariate stratification prostate detection, commonly estimated using transrectal ultrasound (TRUS). While TRUS provides precise measurements, its invasive nature often compromises patient comfort. Transabdominal (TAUS) non-invasive alternative but faces challenges such as lower image quality, complex interpretation, reliance on operator expertise. This study introduces new deep-learning-based framework automatic PV estimation TAUS, emphasizing potential to enable A dataset of TAUS videos from 100 individual patients was curated, with manually delineated boundaries calculated diameters by an expert clinician ground truth. The introduced integrates deep-learning models segmentation both axial sagittal planes, diameter estimation, calculation. Segmentation performance evaluated Dice correlation coefficient (%) Hausdorff distance (mm). Framework's capabilities were volumetric error (mL). demonstrates that it can estimate mean -5.5 mL, which results average relative between 5 15%. images, utilizing deep learning segmentation, shows promising results. It effectively segments the estimates volume, offering reliable, detection.