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

Area of Science:

• Crystallography
• Detector Physics
• Machine Learning

Background:

• X-ray crystallography advancements rely on high-intensity sources and fast detectors.
• Automatic data analysis algorithms are susceptible to detector noise and imperfections.
• Accurate identification of defective pixels is crucial for reliable data processing.

Purpose of the Study:

• To introduce a machine learning-based methodology and program, robust mask maker (RMM), for generating bad-pixel masks.
• To improve the reliability of data analysis in X-ray crystallography by addressing detector imperfections.
• To reduce data storage costs by filtering out uninformative data.

Main Methods:

• Developed a robust mask maker (RMM) program using machine learning and robust statistics.
• Discriminated between normal and abnormal pixels by analyzing X-ray illuminated and non-illuminated measurements.
• Applied RMM to generate bad-pixel masks for large-area X-ray pixel detectors.

Main Results:

• RMM effectively identifies defective pixels, generating accurate bad-pixel masks.
• The method prevents peak finders from misidentifying bad pixels as Bragg peaks.
• Demonstrated improved performance of peak finders and reduced computational load for indexing methods.

Conclusions:

• RMM significantly enhances the reliability of X-ray diffraction data analysis.
• The methodology reduces the storage of uninformative data sets.
• Robust mask generation is essential for efficient and accurate crystallographic studies.