Use this URL to cite or link to this record in EThOS:
Title: Machine learning to assess the fetal brain from ultrasound images
Author: Namburete, Ana Ineyda Luisa
ISNI:       0000 0004 6346 7388
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Obstetric care decisions fundamentally rely upon accurate estimation of gestational age (GA). Ultrasound- (US) based measurements provide reliable estimates of GA, if performed early in pregnancy. However, in low-income settings, the lack of appropriately trained sonographers and the tendency for women to present for care late in pregnancy are barriers to the use of US for dating purposes. In this thesis, we propose to exploit sonographic image patterns associated with dynamic fetal brain development to predict GA. We designed an algorithm which automatically estimates GA from an US scan collected from a single visit, thereby enabling clinically useful estimates of GA to be made even in the third trimester of pregnancy: a period complicated by biological variation and unreliable size-based estimates. The presented model was conceived on the basis that fetal brain development follows a precise spatiotemporal pattern, with folds emerging and disappearing on the surface of the brain (cerebral cortex) at fixed time points during pregnancy. This timing is so precise that post-mortem neuroanatomical and MRI evidence suggest that the 'developmental maturation' of the fetal brain may be a better predictor for GA than traditional size-based estimates. We capitalize on these age-related patterns to develop, for the first time, a unified model which combines sonographic image features and clinical measurements to predict GA and brain maturation. The framework benefits from a manifold surface representation of the fetal head which delineates the inner skull boundary and serves as a common coordinate system based on cranial position. This allows for fast and efficient sampling of anatomically-corresponding brain regions to achieve like-for-like structural comparison of different developmental stages. Bespoke features capture neurosonographic patterns in 3D images, and using a regression forest classifier, we characterize structural brain development both spatially and temporally to capture the natural variation existing in a healthy population (n=448) over an age range of active brain maturation (18 to 34 weeks). Our GA prediction results on a high-risk clinical dataset (n=187) strongly correlate with true GA (r=0.98, accurate within ± 6.10 days), confirming the link between maturational progression and neurosonographic activity observable across gestation. Our model also outperforms current clinical methods, particularly in the third trimester. Through feature selection, the model successfully identified regional biomarkers of neurodevelopmental progression over gestation. Guided by these regions, we present a novel approach for defining and testing hypotheses associated with neuropathological deviations.
Supervisor: Noble, J. Alison Sponsor: Commonwealth Scholarship
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available