ML-ED

Increased spatial resolution in electron detectors through machine learning

Started
September 1, 2022
Status
In Progress
Share this project

Abstract

Hybrid pixel detectors show great potential for use in Electron Microscopy, but poor spatial resolution - caused by large pixel size and multiple scattering in the sensor layer - are holding them back. The goal of this project is to use Machine Learning trained on simulation and measurement data to improve spatial resolution by determining the point where the electron hit the detector and assigning the signal to the correct pixel. This could enable the use of fast, radiation hard hybrid pixel detectors for diffraction experiments over a wide range of energies as well as extend the application to cryo-electron microscopy imaging of biological samples at ≤100 keV resulting as a combined diffraction/imaging detector. If successful, the models developed in this project would not be limited to the ones developed by PSI, but could be used for a wide range of hybrid pixel detectors.

People

Collaborators

SDSC Team:
Luis Barba Flores
Carl Remlinger
Benjamin Béjar Haro

PI | Partners:

PSI, Electron Microscopy and Diffraction group:

  • Dr. Elisabeth Müller
  • Dr. Emiliya Poghosyan
  • Dr. Erik Fröjdh
  • Dr. Xie Xiangyu

More info

description

Motivation

Hybrid pixel detectors have transformed data acquisition of synchrotrons and X-ray Free Electron Lasers and are good candidates for electron microscopy as well. However, at higher energies the incoming electron does not deposit all its energy in one pixel, but rather creates a track spanning several pixels (Figure 1) which reduces the spatial resolution.

Proposed Approach / Solution

The goal of the ML-ED project is to overcome the lack of resolution of HPDs using Machine Learning by identifying the impact points of the electrons on the sensor with pixel- or, for low energies, even sub-pixel resolution.

Figure 1: Side and top view of 200 keV simulated electron tracks in a Silicon sensor. The grid outlines the 25 um pixels of MÖNCH. Each color represents one track. All electrons impinging at x=0, y=0.

After hitting the silicon detector, the electrons scatter through the material, following a path that, although random, exhibits a learnable structure. To leverage this, we propose using convolutional neural networks (CNNs) to predict the electron paths from the images produced by the sensor (Figure 2). This neural network, combined with physics-based models that enforce prediction consistency, enables us to enhance the sensor’s resolution and achieve sub-pixel accuracy.

Impact

The increase in resolution benefits all currently available HPDs due to the shared nature of charge deposition and transport. There are especially applications in cryo-EM imaging of biological molecules and electron diffraction.

Figure 2: Left: Simulated detector response with the electron track causing the signal overlay. Right: Desired detector response after data processing: all signal assigned to the pixel where the electron hit. We also predict the entry point with sub pixel resolution.

Gallery

Annexe

Publications

Additional resources

Bibliography

  • J. P. van Schayck, E. van Genderen, E. Maddox, L. Roussel, H. Boulanger, E. Fröjdh, P. J Peters, R. B. Ravelli  (2020). Sub-pixel electron detection using a convolutional neural network. Ultramicroscopy, 218, 113091. https://www.sciencedirect.com/science/article/pii/S0304399120302424
  • S. Cartier, A. Bergamaschi, R. Dinapoli, D. Greiffenberg, I. Johnson, J. H. Jungmann, D. Mezza, A. Mozzanica, B. Schmitt, X. Shi. Micron resolution of MÖNCH and GOTTHARD, small pitch charge integrating detectors with single photon sensitivity, Journal of Instrumentation, Volume 9, May 2014. https://iopscience.iop.org/article/10.1088/1748-0221/9/05/C05027
  • A. Bergamaschi, M. Andrä, R. Barten, C. Borca, M. Brückner, S. Chiriotti, R. Dinapoli, E. Fröjdh, D. Greiffenberg, T. Huthwelker, A. Kleibert, M. Langer, M. Lebugle, C. Lopez-Cuenca, D. Mezza, A. Mozzanica, J. Raabe, S. Redford, C. Ruder, V. Scagnoli, B. Schmitt, X. Shi, U. Staub, D. Thattil, G. Tinti, C. F. Vaz, S. Vetter, J. Vila-Comamala & J. Zhang. The MÖNCH Detector for Soft X-ray, High-Resolution, and Energy Resolved Applications, Synchrotron Radiation News, Volume 31, 2018 - Issue 6. https://www.tandfonline.com/doi/full/10.1080/08940886.2018.1528428
  • S. Cartier, M. Kagias, A. Bergamaschi, Z. Wang, R. Dinapoli, A. Mozzanica, M. Ramilli, B. Schmitt, M. Brückner, E. Fröjdh, D. Greiffenberg, D. Mayilyan, D. Mezza, S. Redford, C. Ruder, L. Schädler, X. Shi, D. Thattil, G. Tinti, J. Zhang, M. Stampanoni. Micrometer-resolution imaging using MÖNCH: towards G2-less grating interferometry, Journal of Synchrotron Radiation 2016 Nov 1;23(Pt 6):1462-1473. Micrometer-resolution imaging using MÖNCH: towards G2-less grating interferometry
  • L. M. Lohse, M. Vassholz, M. Töpperwien, T. Jentschke, A. Bergamaschi, S. Chiriotti, and T. Salditt. Spectral µCT with an energy resolving and interpolating pixel detector, Optics Express Vol. 28, Issue 7, pp. 9842-9859 (2020). https://opg.optica.org/oe/fulltext.cfm?uri=oe-28-7-9842

Publications

Related Pages

More projects

ML-L3DNDT

Completed
Robust and scalable Machine Learning algorithms for Laue 3-Dimensional Neutron Diffraction Tomography
Big Science Data

BioDetect

Completed
Deep Learning for Biodiversity Detection and Classification
Energy, Climate & Environment

IRMA

In Progress
Interpretable and Robust Machine Learning for Mobility Analysis
No items found.

FLBI

In Progress
Feature Learning for Bayesian Inference
No items found.

News

Latest news

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data
May 1, 2024

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data

We’ve developed a smart solution for wind tunnel testing that learns as it works, providing accurate results faster. It provides an accurate mean flow field and turbulence field reconstruction while shortening the sampling time.
The Promise of AI in Pharmaceutical Manufacturing
April 22, 2024

The Promise of AI in Pharmaceutical Manufacturing

The Promise of AI in Pharmaceutical Manufacturing

Innovation in pharmaceutical manufacturing raises key questions: How will AI change our operations? What does this mean for the skills of our workforce? How will it reshape our collaborative efforts? And crucially, how can we fully leverage these changes?
Efficient and scalable graph generation through iterative local expansion
March 20, 2024

Efficient and scalable graph generation through iterative local expansion

Efficient and scalable graph generation through iterative local expansion

Have you ever considered the complexity of generating large-scale, intricate graphs akin to those that represent the vast relational structures of our world? Our research introduces a pioneering approach to graph generation that tackles the scalability and complexity of creating such expansive, real-world graphs.

Contact us

Let’s talk Data Science

Do you need our services or expertise?
Contact us for your next Data Science project!