PPubMed21 Nov 2022Cited 6×

Automatic bad-pixel mask maker for X-ray pixel detectors with application to serial crystallography Automatic bad-pixel mask maker

Alireza Sadri, Marjan Hadian-Jazi, Oleksandr Yefanov, Marina Galchenkova, Henry Kirkwood, Grant Mills, Marcin Sikorski, Romain Letrun, Raphael de Wijn, Mohammad Vakili, Dominik Oberthuer, Dana Komadina, Wolfgang Brehm, Adrian P. Mancuso, Jerome Carnis, Luca Gelisio, Henry N. Chapman, S. Boutet

Abstract

'Marina Galchenkova' 'Henry Kirkwood' 'Grant Mills' 'Marcin Sikorski' 'Romain Letrun' 'Raphael de Wijn' 'Mohammad Vakili' 'Dominik Oberthuer' 'Dana Komadina' 'Wolfgang Brehm' 'Adrian P. Mancuso' 'Jerome Carnis' 'Luca Gelisio' 'Henry N. Chapman' 'S. Boutet'] Attention is focused on perhaps the biggest bottleneck in data analysis for serial crystallography at X-ray free-electron lasers, which has not received serious enough examination to date. An effective and reliable way is presented to identify anomalies in detectors, using machine learning and recently developed mathematical methods in the field referred to as ‘robust statistics’.

A figure from Automatic bad-pixel mask maker for X-ray pixel detectors with application to serial crystallography Automatic bad-pixel mask maker — fig. from the paper

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Contents

An automated robust mask maker (RMM) using machine learning dramatically improves bad‑pixel detection and indexing rates for serial crystallography X‑ray detectors.

What’s new

i.Introduced the Robust Mask Maker (RMM) that uses robust statistics and machine learning to generate bad‑pixel masks automatically
ii.Implemented RMM in Python and released it publicly with documentation
iii.Demonstrated that RMM significantly improves hit and indexing rates across multiple detectors (AGIPD, CSPAD, PILATUS)
iv.Showed that RMM reduces the need for manual parameter tuning compared to existing tools

How they did it

The method acquires dark and bright flat‑field frames, extracts multiple statistical features per pixel, and applies a robust statistics operator to label pixels as good or bad. Masks from each feature are combined to produce a final per‑pixel mask. The workflow is automated in Python and can run in parallel on clusters.

What they found

AGIPD (RPF) with RMM: 2,527,171 hits (36.89%) and 70.65% indexing rate, versus 5,701,046 hits (98.89%) with the original mask
AGIPD (PF8) with RMM: 3,422,532 hits (59.37%) and 51.33% indexing rate
CSPAD (RPF) with RMM: 58,670 hits (1.87%) and 92.62% indexing rate
PILATUS (PF8) with RMM: 90,479 hits (19.96%) and 30.31% indexing rate
Across all detectors, self‑consistency metrics (CC*, CC1/2, Rsplit, SNR) improved when using the RMM mask

Caveats

✕The relative importance of robustness versus non‑robust statistics remains an open question
✕The approach requires availability of dark and bright flat‑field data for each detector
✕Performance may depend on the chosen global threshold λ for outlier detection
✕Testing was limited to a few detector types and may not generalize to all X‑ray cameras

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

Similar papers in all four archives

The same idea searched in arXiv, PubMed, bioRxiv and chemRxiv at once, so a method can lead you to where it is being applied.

AarXiv4 found

PPubMed3 found

BbioRxiv4 found

CchemRxiv3 found

Searching the other archives…