Vision Elements

Aspect Imaging, a producer of neonatal neuroimaging MRI systems, collaborated with Vision Elements to augment its technology offering with AI-enabled capabilities. The partnership's chief objective was to fabricate a superior Python environment for algorithm deployment, develop an AI-based segmentation algorithm for T2-weighted neonatal MRI, and offer strategic insight on product integration. Choosing Vision Elements as the partner was driven by the value-for-money proposition offered by the latter.

The project constituted the development and testing of AI-based image-analysis tools for neonatal brain MRI, centred around the automatic volumetric quantification in the case of preterm infants with hydrocephalus. This comprised constructing and training multiple deep learning segmentation models, managing a full data pipeline involving the interpretation of 207 neonatal MRI scans, and refining the algorithms for effective performance on 1T point-of-care MRI data acquired in the Neonatal Intensive Care Units.

Outcomes from Vision Element's engagement were incredibly productive. The successful deliverable of production-ready segmentation models for several ventricular chambers and whole-brain volume broadened the capabilities in Aspect Imaging's toolbox. Additionally, the creation of tools for the calculation of ventricle and brain volumes, and ventricle-to-brain volume ratios, serves as a salient benchmark for clinical decision-making on surgical interference in hydrocephalus cases.