Preterm birth (PTB) is the leading cause of neonatal morbidity and mortality. It occurs in 5-18% of births, with approximately 15,000,000 PTBs per year.
Currently, predicting PTB is based on identifying a short cervix via transvaginal ultrasound. Unfortunately, this methodology has a sensitivity of ≤50%. Many image-based methodologies have been reported, but they do not consistently perform as well as cervical length.
The machine learning approach used in this project focuses on extracting important morphological features from the ultrasound images and using the features to build nonlinear models that predict PTB better than CL.
David Bustamante, Y. Yan, Maryam Basij, Azin Gelareh, Edgar Hernandez-Andrade, Seyedmohammad Shams, Mohammad Mehrmohammadi, "Cervix Ultrasound Texture Analysis to Differentiate Between Term and Preterm Birth Pregnancy: A Machine Learning Approach," in IEEE IUS 2022, Venice, Italy, 2022, pp. 1-4.