Lawrence H. Staib

Timeline

1. 1982–1991

   Cornell physics, Yale doctorate

   He earns an AB in physics at Cornell University in 1982 and a doctorate in engineering and applied science at Yale University in 1990. A postdoctoral fellowship at Yale School of Medicine follows in 1991.

2. 1989–1994

   Deformable models for boundary finding

   With James Duncan, he presents Parametrically Deformable Contour Models at CVPR in 1989, a probabilistic deformable model of object boundaries built on the elliptic Fourier decomposition and solved as an optimization problem. In 1994, he coauthors Deformable Boundary Finding Influenced by Region Homogeneity with Amit Chakraborty and Duncan, a CVPR paper that uses Green's theorem to derive the boundary of a homogeneous region and integrates it with a gradient-based boundary finder.

3. 1998

   Statistical shape models at CVPR

   In 1998, he and Yongmei Wang present Boundary Finding with Correspondence Using Statistical Shape Models at CVPR. The Bayesian approach models objects with global shape parameters derived from the statistical variation of boundary points in a training set, and the experiments use synthetic and real medical images of the brain and heart.

4. 2015

   An AIMBE fellowship

   In March 2015, AIMBE announces his induction into its College of Fellows, Class of 2015, for outstanding contributions to model-based medical image analysis. He is then a professor of biomedical engineering, diagnostic radiology and electrical engineering at Yale University.

5. 2022

   A MICCAI fellowship

   In October 2022, Yale announces that he is elected a Fellow of the Medical Image Computing and Computer Assisted Intervention Society (MICCAI) for significant scientific contributions to medical image computing and analysis.

6. 2025

   Adaptive dropout with Rate-In

   In 2025, he coauthors Rate-In with Tal Zeevi, Ravid Shwartz-Ziv, Yann LeCun and John Onofrey, accepted to the CVPR 2025 conference. The algorithm adjusts dropout rates during inference by measuring the information loss dropout causes in each layer's feature maps, which improves calibration and sharpens uncertainty estimates on synthetic data and medical imaging tasks.

7. 2026

   Self-supervised learning on fMRI

   In 2026, he is a coauthor of Learning robust and task-invariant functional representation from fMRI through Siamese self-supervised learning, published in Medical Image Analysis.

8. As of October 2026

   Professor and undergraduate director

   He is director of undergraduate studies for biomedical engineering in the Yale College Programs of Study 2026–2027, and a professor of radiology and biomedical imaging, of biomedical engineering and of electrical engineering at Yale. He is Associate Editor of IEEE Transactions on Biomedical Engineering and a member of the editorial board of Medical Image Analysis. His research develops statistical and geometric deformable models for segmentation, along with machine-learning and model-based methods for cancer, neuroimaging and cardiology.