Thomas Hinano Keller

PhD Researcher · AI for Medical Imaging

I build AI systems for medical imaging and digital pathology. Currently pursuing my PhD, focused on data-efficient methods for clinical decision support in low-resource settings.

Projects

AIMIX – AI for Maternal & Fetal Health

Active

ERC-funded project developing robust, data-efficient AI for fetal ultrasound analysis in low-resource African settings — improving prenatal care and clinical decision support.

Digital Pathology with Foundation Models

Conference

AI pipelines for computational pathology using whole slide images. Leveraged Gigapath and Canvoi foundation models to classify BMI groups and predict HER2 status from histopathology embeddings across multiple magnifications.

COPD Severity Staging via U-Net

U-Net based segmentation for automated classification of COPD severity from medical imaging, supporting clinical decision-making.

↗ GitHub

Deep Neural Network in C

From-scratch neural network in C with parallelized training via OpenMP and CUDA.

↗ GitHub

Perona-Malik with Mimetic Method

Anisotropic diffusion for image denoising using the Mimetic Method — preserves structural features while reducing noise.

↗ GitHub

Consulting

I help companies explore and apply AI in their daily operations — from understanding what's possible to designing and implementing machine learning workflows. Background in data science, ML research, and building scalable AI systems.

Book a call →