A Nigerian researcher based in the United States has developed an artificial intelligence-powered system capable of detecting and mapping brain tumours without relying on the expensive computing hardware that has traditionally limited the adoption of AI in healthcare.
The innovation, created by Oladele Ayomide Benjamen, is expected to improve access to AI-assisted brain tumour diagnosis in hospitals, universities and research institutions operating with limited technological resources, particularly across developing countries.
Benjamen, a doctoral student in Chemistry at Southern Methodist University, Texas, designed the lightweight AI model to function efficiently on standard desktop and laptop computers using Central Processing Units (CPUs), eliminating the need for costly Graphics Processing Units (GPUs) that power most advanced medical AI systems.
His research, titled "Lightweight 3D U-Net for Brain Tumor Segmentation on CPUs: Enabling Deep Learning in Low-Resource Environments," demonstrates that sophisticated deep-learning models can achieve accurate brain tumour segmentation while running on conventional computing systems already available in many healthcare facilities.
Explaining the motivation behind the project, Benjamen said the goal was to ensure that hospitals with limited resources are not excluded from benefiting from advances in artificial intelligence.
"My research is helping make medical AI more accessible, so that hospitals and researchers with limited resources are not left behind," he said.
The development comes at a time when artificial intelligence is increasingly transforming cancer diagnosis worldwide. A 2025 systematic review and meta-analysis covering 79 independent studies reported that AI systems recorded an overall diagnostic accuracy of about 95.2 per cent in detecting and classifying brain tumours from MRI scans, highlighting the growing role of machine learning in modern medicine.
While many of these AI-powered diagnostic systems require high-performance GPU-equipped computers, Benjamen's model has been specifically optimised for CPUs, making it significantly more affordable and practical for healthcare providers in resource-constrained settings.
The researcher, who earned a Bachelor's degree in Chemistry from Obafemi Awolowo University, Ile-Ife, previously worked as a Biomedical AI Engineer at the Medical Artificial Intelligence (MAI) Laboratory in Lagos. He also held artificial intelligence and bioinformatics positions at DycoVue and Genomac before beginning his doctoral studies in the United States.
The CPU-based brain tumour detection project was developed at MAI Lab alongside Charity Umoren and other researchers under the supervision of the laboratory's Executive Director, Dr. Maruf Adewole. The research also benefited from contributions by Prof. Fatade and Scientific Director at McGill University, Canada, Prof. Udunna.
According to Benjamen, the research addresses one of the biggest barriers facing healthcare systems across sub-Saharan Africa, where access to MRI facilities remains critically low. He noted that fewer than one MRI scanner is available for every one million people across many parts of the region. Even where MRI scanners exist, hospitals often lack the expensive computing infrastructure required to deploy advanced AI-assisted diagnostic tools.
His lightweight deep-learning model is designed to remove that obstacle by allowing healthcare providers to analyse MRI images for brain tumours using existing computers, reducing operational costs while expanding access to advanced diagnostic technology.
Beyond the CPU-powered innovation, Benjamen also contributed to MAPS-Glioma, an international research initiative involving scientists from Nigeria and Tanzania focused on improving AI-assisted brain tumour analysis using medical imaging collected from resource-limited environments.
The project formed part of the internationally recognised BraTS-Africa Challenge, where the research team ranked among the competition's top-performing groups, receiving recognition for the quality, innovation and practical relevance of their work.
Looking beyond his current research, Benjamen said his ambition is to integrate artificial intelligence with chemistry, nanotechnology and medical imaging to develop practical healthcare solutions capable of improving disease diagnosis, treatment and patient outcomes.
"My long-term goal is to combine technology and science to create solutions that are not only innovative but useful. By bringing together AI, chemistry and healthcare research, I hope to develop tools that improve lives, support medical progress and solve real societal problems," he said.
His doctoral work has also expanded into nanochemistry, where he is investigating nanocluster-based platforms that could contribute to future breakthroughs in cancer diagnosis, targeted therapy and biomedical materials.
Commenting on the significance of the innovation, Raymond Confidence, Programme Chair of Spark Academy at McGill University, Canada, described the research as an important step toward making AI-powered healthcare more accessible across low-resource countries.
According to him, the technology has the potential to help hospitals in Nigeria and other developing nations deploy deep-learning solutions using standard computers, reducing dependence on expensive infrastructure while strengthening local capacity for AI-driven healthcare delivery.
Medical experts, however, continue to emphasise that while artificial intelligence is rapidly improving cancer detection and diagnostic accuracy, it is intended to complement—not replace—the expertise of healthcare professionals. Final diagnosis, treatment planning and clinical decisions remain the responsibility of qualified medical practitioners.
Benjamen's latest achievement adds to the growing list of contributions by Nigerian researchers advancing artificial intelligence in medicine. It also reinforces ongoing efforts to develop affordable technologies capable of expanding access to quality healthcare for underserved populations across Africa and other low-resource regions.