about
Smriti Joshi
ML Researcher · Trustworthy AI in Healthcare
PhD Researcher at BCN-AIM, Universitat de Barcelona
I'm a PhD Researcher at BCN-AIM, Universitat de Barcelona, working on trustworthy AI for medical imaging — models that hold up under domain shift and multi-center clinical data, rather than just in-distribution benchmarks. My work sits at the intersection of domain adaptation, uncertainty estimation, and generative imaging, mostly applied to breast MRI within the Horizon Europe project RadioVal.
A good part of my time also goes into building shared infrastructure for the field: I co-organize community benchmarks like MAMA-MIA and MAMA-SYNTH, and help coordinate AI schools and mentorship programs across Africa and South Asia through AFRICAI.
Selected Publications
- Dense Temporal Contrast Synthesis via Conditioned Latent Transport2026Submitted to Medical Image Analysis
- The MAMA-MIA Challenge: Advancing Generalizability and Fairness in Breast MRI Tumor Segmentation and Treatment Response Prediction2025Equal contribution (Garrucho, Joshi). Submitted to IEEE Transactions on Medical Imaging
- Single Image Test-Time Adaptation via Multi-View Co-TrainingIn International Conference on Medical Image Computing and Computer Assisted Intervention, 2025
- Stakeholder Engagement for Trustworthy AI: Experiences and Results from the RadioVal Project on Breast Cancer ImagingIn Trustworthy AI in Cancer Imaging Research, 2025
- Leveraging epistemic uncertainty to improve tumour segmentation in breast MRI: an exploratory analysisIn Medical Imaging 2024: Image Processing, 2024
-
nn-UNet training on CycleGAN-translated images for cross-modal domain adaptation in biomedical imagingIn International MICCAI Brainlesion Workshop, 2021