This project is undertaken by a master’s student who is passionate about leveraging artificial intelligence to address critical gaps in healthcare with Sub-Saharan Africa.
The shortage of neuro-oncologists and neurosurgeons in Sub-Saharan Africa necessitates innovative solutions to alleviate their workload and improve brain tumour diagnosis. AI-driven tools, such as segmentation models, could enhance diagnostic accuracy and efficiency. However, the development of these models within the region is hindered by the lack of locally available datasets, largely due to the high cost of data annotation.
Applying models developed in Western contexts poses challenges, as these models often struggle with domain shift and may fail to generalise to African medical images, which have distinct characteristics. Rachel is eager to explore Test-Time Adaptation (TTA) techniques, which allow AI models to dynamically adjust to unseen data during inference without the need for extensive retraining
