article · Scientific Reports
Artificial intelligence tools for medical imaging are predominantly developed in high-income countries, raising questions about their reliability in resource-constrained environments. This study evaluates how well a deep learning model trained to identify standard fetal ultrasound planes generalises to clinical settings with limited imaging infrastructure. An initial classifier was trained on data from 1,792 patients in Spain and tested under optimal conditions on 1,008 patients in Denmark. To address the performance gap when applied to low-resource settings, researchers tested transfer learning and domain adaptation strategies using small datasets of 25 patients each from five African countries: Egypt, Algeria, Uganda, Ghana, and Malawi. The optimisation framework successfully adapted the high-resource model to heterogeneous African clinical environments, substantially increasing recall while preserving high precision despite the small local sample sizes.
Sub-Saharan Africa experiences high perinatal mortality rates partly because access to antenatal screening is constrained. Artificial intelligence can assist healthcare providers in capturing correct fetal ultrasound planes for abnormality diagnosis. Adapting existing global models ensures that clinics with limited data and basic equipment can benefit from diagnostic AI without needing massive, expensive local datasets.
This research provides applied and tested algorithms suitable for healthcare technology developers building fetal ultrasound guidance software. The target end-users are clinical workers in low-resource maternity wards who require support in capturing standard diagnostic planes. While evaluated on clinical datasets from five African nations, the approach remains at an experimental validation stage requiring larger trials and regulatory clearance before clinical integration.
AI-generated from the published abstract. Always read the original work before citing.
Most artificial intelligence (AI) research and innovations have concentrated in high-income countries, where imaging data, IT infrastructures and clinical expertise are plentiful. However, slower progress has been made in limited-resource environments where medical imaging is needed. For example, in Sub-Saharan Africa, the rate of perinatal mortality is very high due to limited access to antenatal screening. In these countries, AI models could be implemented to help clinicians acquire fetal ultrasound planes for the diagnosis of fetal abnormalities. So far, deep learning models have been proposed to identify standard fetal planes, but there is no evidence of their ability to generalise in centres with low resources, i.e. with limited access to high-end ultrasound equipment and ultrasound data. This work investigates for the first time different strategies to reduce the domain-shift effect arising from a fetal plane classification model trained on one clinical centre with high-resource settings and transferred to a new centre with low-resource settings. To that end, a classifier trained with 1792 patients from Spain is first evaluated on a new centre in Denmark in optimal conditions with 1008 patients and is later optimised to reach the same performance in five African centres (Egypt, Algeria, Uganda, Ghana and Malawi) with 25 patients each. The results show that a transfer learning approach for domain adaptation can be a solution to integrate small-size African samples with existing large-scale databases in developed countries. In particular, the model can be re-aligned and optimised to boost the performance on African populations by increasing the recall to [Formula: see text] and at the same time maintaining a high precision across centres. This framework shows promise for building new AI models generalisable across clinical centres with limited data acquired in challenging and heterogeneous conditions and calls for further research to develop new solutions for the usability of AI in countries with fewer resources and, consequently, in higher need of clinical support.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1038/s41598-023-29490-3
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
