Research · Applied ML for health
I work on computer vision for medical images collected in low-resource settings: brain MRI, blood-film microscopy and retinal scans. The common thread is getting reliable results from small, local datasets, and being honest about where strong baselines are enough.
Research themes
Each theme maps to at least one paper, poster or active project below.
How well do standard pipelines like nnU-Net hold up on limited Sub-Saharan glioma MRI data?
YOLO and Vision Transformers for leukemia detection, and AI for malaria parasite detection and classification.
Translating pose sequences directly to text for continuous sign language recognition.
Automated detection of diabetic retinopathy from retinal images to support screening.
Publications
My name is underlined in each author list. Use "Copy citation" for a ready-made reference.
Segmentation, Classification, and Synthesis for Brain Tumors and Traumatic Brain Injuries. MICCAI 2025. Lecture Notes in Computer Science, vol. 16376. Springer, Cham.
Evaluates glioma segmentation approaches on a limited MRI cohort from Sub-Saharan Africa and makes the case for nnU-Net as the baseline new methods should beat. Presented in Daejeon, Republic of Korea, September 2025.
Lowami, U., Stephen, A. B., Raymond, C., Zhang, D., Adewole, M., Anazodo, U. C., Kurt, M., Olatunji, D., & Atabonfack, B. L. Reassessing Glioma Segmentation Strategies: nnU-Net as a Strong Baseline on Limited Sub-Saharan MRI Data. In Segmentation, Classification, and Synthesis for Brain Tumors and Traumatic Brain Injuries (MICCAI 2025). Lecture Notes in Computer Science, vol 16376. Springer, Cham. https://doi.org/10.1007/978-3-032-16365-3_30
IEEE AFRICON 2025, Polokwane, South Africa.
Combines YOLO object detection with Vision Transformer classification to flag leukemia in microscopic blood-film images, aiming for a pipeline that is practical for labs with limited specialist capacity.
Stephen, A. B., Olotuche, M. J., Aman, M., & Biyabani, A. A. (2025). Automated and Accessible Leukemia Detection from Microscopic Blood Films Using YOLO and Vision Transformers. IEEE AFRICON 2025, Polokwane, South Africa.
Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, MSLR 2025.
Translates body-pose sequences directly into text for continuous sign language recognition.
Johnny, S., Guda, B., Stephen, A. B., & Gueye, A. (2025). AutoSign: Direct Pose-to-Text Translation for Continuous Sign Language Recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, MSLR.
AI in Health Africa Conference 2025, Makerere AI Health Lab.
Poster on automated detection of diabetic retinopathy from retinal images as a screening aid.
Alpha, M., Francois, M., & Stephen, A. B. (2025). Automated Diabetic Retinopathy Detection. Poster, AI in Health Africa Conference 2025.
Ongoing work
Active research roles and projects that have not been published yet.
Research Assistant
Graduate Research Assistant
Talks & events
Where I have presented work or joined research communities.
Paper presentation: leukemia detection from blood films with YOLO and Vision Transformers.
Paper presentation: glioma segmentation on limited Sub-Saharan MRI data.
Participant in a programme on technology, ethics and society.
Attended the inaugural NEISA conference.
Collaborate
I'm happy to talk about datasets, annotation, evaluation or joint papers.