Henry Pinkard

Google Scholar

GitHub

hbp [at] berkeley [dot] edu

I’m a postdoctoral researcher in the Computational Imaging Lab at UC Berkeley, within the department of Electrical Engineering and Computer Sciences and the Berkeley AI Research Lab. I received my PhD (Computational Biology) and MS (EECS) here, advised by Laura Waller.

My research focuses on using information theory and machine learning to design better imaging systems. Before coming to Berkeley, I worked in the UCSF Biological Imaging Development Center, where I did immunology research and built and programmed microscopes. I also helped develop Micro-Manager, an open-source software for the control of microscopes.





Research highlights

Information-theoretic design

Information-driven design of imaging systems — A method to measure and optimize how much information imaging systems capture, applicable to diverse systems ranging from consumer cameras to radio telescopes observing black holes.
(website) (pre-print) (code)
Extended channel A visual introduction to information theory — A practical introduction to the fundamentals of information theory, describing concepts such as data compression and accurate transmission of messages in the presence of noise.
(paper) (code+figures)

Computational microscopy

The Berkeley single-cell computational microscopy (BSCCM) dataset — contains over >400k images of white blood cells under varied LED illumination patterns, paired with protein expression measurements. It provides standardized training data for computational imaging and vision algorithms with biomedical applications.
(website)
adaptive_deep_generative_model_microscope Microscopes are coming for your job — Speculating about the future possibilities agentic artificial intelligence and reinforcement learning in microscopy.
(paper)
Learned adaptive multiphoton illumination microscopy — A technique where a neural network dynamically adjusts multiphoton microscope laser power during scanning, enabling immune cell imaging at previously impossible scales. This allows observation of T cell and dendritic cell organization during early immune responses.
(paper) (tutorial) (data)
Deep learning for single-shot autofocus microscopy — A fast autofocusing technique combining custom illumination patterns with a physics-based neural network architecture to predict focus corrections from single images. Requires far fewer parameters than standard networks while maintaining accuracy.
(paper) (tutorial) (code)

Open-source software

Pycro-Manager banner Pycro-Manager — An open-source Python package for controlling microscopes, enabling automated experiments and real-time adaptive imaging. Works with hundreds of microscope components and handles large-scale data acquisition.
(documentation) (paper) (code)
Micro-Magellan — A microscopy acquisition software for defining and imaging arbitrary three-dimensional regions in large samples like tissue slides and multi-well plates. Enables simultaneous graphical control and Python-based image processing during acquisition.
(documentation) (paper) (code)