Use case "Pre-screening of diffraction images"

From EdnaWiki
Jump to: navigation, search

Summary

This use case describes the necessary fuctions for the pre-screening module. In DNA 1.X only the images used for indexing were prescreened but for DNA 2.0 all images should be pre-screened.

Importance Very important
Priority High priority
Use Frequency Frequently used/ Always ???
Direct Actors Kernel
Pre-requirements Kernel running

Input

Required

If more than one image, all images must be of the same image type.

A list of one or several reference diffraction images:

  • Data collection parameters for each image:
    • Path to image directory
    • Image file name

Output

Required

  • A list of found spots. For each spot:
    • Intensity of spot (box integration)
    • I/sigma
    • Resolution
    • Coordinates
  • A list of diffraction rings. For each ring:
    • Intensity
    • I/sigma
    • Width
    • Resolution
    • Center of ring ?( this gives beam position )
  • Comment if image is blank or weak spots

Optional

  • Threshold for spot finding during indexing

Main Success Scenario

The pre-screening module loads each diffraction image and pre-screens them. The pre-screening for each image consists of the following actions:

Read the image

The pre-screening module must be able to read images of the following types:

  • ADSC (q4, q4-2x, q210, q210-2x, q315, q315-2x)
  • MAR CCD (mar165, mar225, mar325)
  • CBF format

Diffraction ring search

The diffraction ring search algorithm must be able to find diffraction rings even if the provided beam centre is incorrect.

One proposal is to convert the image to polar coordinates and look for intensities which have the same radius:

http://abatools.sourceforge.net/images/PolarCoordinates.png

This assumes of course that the beam centre is correct so this procedure must be repeated for many positions around the provided beam centre.

Diffraction ring integration

  • After integrating the diffraction rings the following information should be available for each diffraction ring:
    • Intensity (I)
    • sigma(I)
    • Beam centre
    • Diffraction ring width

Spot finding

Before spot finding any detected diffraction rings should be removed from the image.

  • The spot finding algorithm should be fast and robust.
  • How are we going to identify a spot? For example, a spot should fulfill the following requirements:
    • Background threshold must be determined
    • Minimum number of pixels with values over a threshold
    • Full width > predefined value (minimum number of pixels)
    • I/sigma(I) > predefined value

Spot integration

  • For each spot, intensity (I) and variation (sigma(I)) should be calculated.
  • How large should the box be for spot integration?
  • How are we going to detect overlapping peaks?

Variations

If one or more images cannot be read an error message should be returned:

  • File not existing,
  • Error in image format
  • File not complete
  • File not readable due to permissions

Notes and Questions

Presence of rings doesn't mean that the pattern is useless.

Rings only - useless Rings + diffraction spots - useful

Presence of strong spots - salt ?