1. Radiation Protection in Digital Radiology
Digital Radiographic Image Processing
L05

2. Educational Objectives
List three main purposes of digital image processing
Explain the term “greyscale histogram”
Show how radiographic technique factors affect the greyscale histogram
Suggest how errors in digital image processing can contribute to unnecessary radiation exposure to patients
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

3. The quality of any monochrome image can be described in conventional terms.
Density (darkness)
Contrast
Sharpness
Noise
Artefacts
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

4. Advantages of DR images vs. analogue images
Density can be modified.
Contrast can be modified.
Sharpness can be modified.
Noise can be modified.
Result of 20 years of development!
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

5. Advantages of analogue images vs. DR images
Density function inherently nice.
Contrast inherently higher.
Sharpness inherently higher.
Noise inherently lower.
Result of 110 years of development!
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

6. DR density is adjustable and arbitrary
Acquisition is independent from display
Code values in the raw DR image can be translated to any display level
This allows DR to compensate for over- and under-exposure, producing a consistent appearance
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

7. Incorrect exposure factor selection in screen-film radiography
Overexposed
Underexposed
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

8. DR compensates for incorrect exposure factor selection
Overexposed
Underexposed
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

9. The raw DR image has low contrast
DR has an extremely wide latitude, which implies low contrast for an imaging system that is “display limited” (limited by the latitude of the display).
DR code values can be remapped to generate high contrast for “values of interest” (VOI), while sacrificing contrast for other values.
This is the primary purpose of DR image processing
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

10. Idealised Greyscale Histogram
Goal is to represent anatomy, A, with good contrast
B is air
C is scatter contribution - only outside collimators
D is scatter contribution -only image of anatomy outside collimated area, barium, or lead
# pixels A B C D
Exposure (or greyscale)
Region of clinical interest
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

11. Increased mAs shifts the Greyscale Histogram
# pixels A B C D
Exposure (or greyscale)
Region of clinical interest
Changing mAs does not affect subject contrast, as long as the dynamic range is not exceeded
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

12. Increased kVp squeezes the Greyscale Histogram
# pixels B A C D
Exposure (or greyscale)
Region of clinical interest
Higher kVp => less subject contrast
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

13. Detection of collimator boundaries or anatomy, “exposure recognition”
Primary job of image processing: identify values of interest and maximize their contrast.
Detection of collimator boundaries or anatomy, “exposure recognition”
Window width and window level are adjusted relative to greyscale histogram
Density is thus also adjusted
This is “acquisition processing”
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

14. Did we skip a step?
How did we assure that the response of the detector was uniform over the entire field of view?
“Pre-acquisition processing”, or “preprocessing”, corrects for detector imperfections and variable response.
Some include auto-ranging in preprocessing.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

15. Detector characteristic function for CR
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

16. Raw data Raw, ranged data
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

17. “Matched latitude” is another feature of histogram re-scaling
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

18. Code values can be remapped in more complex ways to modify contrast
Modifying contrast is the secondary role for image processing
Contrast is compromised for some values of interest in order to enhance contrast in others
“post-acquisition processing”
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

19. There are many brand-names for post-acquisition processing
Look-up-table (LUT, Gradation Processing)
Unsharp Mask (Frequency Processing)
Multi-frequency Processing (Musica®)
Multi-Objective Frequency Processing
Dynamic Range Control
Tomographic Artifact Suppression
Energy Subtraction (Dual Energy)
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

20. Code values can be remapped to a non-linear function
This function might have lower contrast for lighter and darker features with higher contrast for values in the middle range, to achieve a film-like appearance.
Code values within the values of interest are translated by means of a Look-up Table (LUT).
This is “Gradation Processing”, “Sensitometry”, and “grey-scale rendition”
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

21. Gradation-processed data
Raw, ranged data
Gradation-processed data
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

22. The DR image has limited sharpness
Sharpness is limited by pixel dimensions
The smallest feature that can be resolved by CR is a “line pair” represented by one “dark” pixel next to a “light” pixel.
Maximum spatial resolution is the sampling rate (pixels per mm) divided by 2 (pixels per line pair)
This is also called the “Nyquist Frequency” or “Nyquist Limit”
A large format cassette with 2000 pixels along the 35 cm dimension would have about 6 pixels per mm and a maximum spatial resolution of 3 lp/mm
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

23. Practical resolution is less than the Nyquist frequency
Factors besides sampling compromise sharpness
X-ray focal spot dimensions
Blur in Indirect DR and CR
Optical and mechanical imprecision in IDR and CR
Afterglow in fast-scan dimension in CR
Limit of resolution is where Modulation Transfer Function (MTF) has decreased to 10%
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

24. Enhancing sharpness: a secondary purpose of image processing
If one can selectively increase the contrast of features in the image that represent large changes in code value over a few pixels, one can increase sharpness.
Two methods
Unsharp Mask (Frequency Processing)
Musica® Edge Contrast
“edge restoration”
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

25. Unsharp mask process Start with original image
Create a blurred version of the original image by averaging all pixels within a small region called a “kernel”.
A large kernel blurs large features
A small kernel blurs small features
Subtract the blurred image from the original image to make a difference image or mask
The mask contains features that were NOT blurred
Add the mask back to the original image
Resulting image has enhanced features that were NOT blurred
Enhancement controlled by a “boost” factor
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

26. Unsharp mask process blurred Original image Blurred image
Orig – blurred
Orig + Diff
Difference image
Sharpened image
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

27. Musica® Edge Contrast
Raw image is decomposed into sub-bands, each representing an octave of spatial frequencies.
Adding all sub-bands together would reconstitute the original image
The contrast of features in each sub-band is modified according to a function.
The degree of enhancement is controlled by the value of a single parameter.
Differential enhancement is controlled by a second parameter value.
The modified sub-bands are added back together to create a modified image.
Extra enhancement of high frequency sub-bands emphasizes edges
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

28. Gradation-processed data
Raw, ranged data
Gradation-processed data
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

29. Musica® Processed data Gradation-processed data
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

30. Dynamic Range Control Original image blurred f(blurred)
Orig + f(blurred)
Blurred image
DRC image
Dynamic Range Control
(actually a form of “contrast enhancement”)
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

31. Two images are acquired
Dual Energy Subtraction Imaging: Uses low energy image and high energy image …
Two images are acquired
Weighted combination and subtraction of these images produces “bone only” and “soft tissue only “ images
Quality of images depends on energy separation
Bone only image
Soft tissue only image
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

32. Conventional vs. DES images
Where is the lesion? Is it calcified?
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

33. Sufficient x-rays must reach the detector to produce the radiographic image.
At the same dose, the smaller the pixel size, the fewer x-rays in each pixel, and the worse the noise.
Larger the pixel size, worse the sharpness.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

34. 9 µGy
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

35. 3 µGy
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

36. 0.9 µGy
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

37. Post-acquisition processing can reduce noise
Post-acquisition processing can reduce noise. (generic term is “noise reduction”)
Because noise is considered as high frequency variation, attenuating high spatial frequencies can reduce noise.
This is effectively a “high pass filter”
Unsharp Mask can do this
Musica® Noise Reduction can do this
This also attenuates small clinical features!
Corollary is that, enhancing small features also enhances noise!
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

38. Optimization in DR imaging cannot ignore patient dose!
In order to make a diagnostic radiographic image, a sufficient number of x-rays must reach the detector.
Unfortunately, the x-rays must pass through the patient to reach the detector.
The ALARA Principle dictates that the examination should be performed with the lowest reasonable dose to the patient.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

39. Acquisition processing involves assumptions:
Radiographic technique
Composition of anatomic region imaged
Use of collimation
Images with very different greyscale histograms
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

40. Post-acquisition processing is controlled by exam-specific parameters
There are literally thousands of permutations of allowable parameter settings.
Extreme values can have dramatic effects on the image.
There is no general agreement on the optimum values for the parameters.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

41. Auxiliary purpose of image processing: improve usability
Imprint demographic overlays
Add annotations
Apply borders or shadow masks
Flip and rotate
Increase magnification
Conjoin images
Scoliosis
Full leg
Modify sequence of views
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

42. Conjoined images: early vs. modern
Note: different contrast
Better contrast matching
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

43. Inappropriate roles for image processing
Compensate for inappropriate radiographic technique
Compensate for poor calibration of acquisition and display
Deletion of non-diagnostic images
Recovery of non-diagnostic images to prevent re-exposure is last resort, not routine activity!
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

44. Bad practice still translates into bad images.
Automation has not been invented to correct for
patient motion
poor inspiration
bad positioning
improper collimation
incorrect alignment of x-ray beam and grid
wrong exam performed
wrong patient examined
double-exposure.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

45. Radiographers need to recognize image artefacts ...
… and take appropriate action when artefacts occur.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

46. Conclusions: DR image quality can be described in conventional terms.
DR image processing has three purposes:
Identify values of interest and maximize their contrast
Modify contrast within values of interest
Improve usability of the image
Image quality cannot be optimized without considering patient dose.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

47. Answer True or False
Acquired images from DR system are independent of display
DR has wide latitude
Spatial resolution of DR images are limited by pixel dimensions
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

48. Answer True or False
True. Acquisition is independent from display, code values in the raw DR image can be translated to any display level
True. DR has extremely wide latitude, ie., it has low contrast and is limited by the latitude of the display
True. The factors involved are focal spot thickness, blur in indirect DR (IDR) and CR, after glow and optical and mechanical imprecision in IDR and CR.
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

49. References
Flynn MJ Processing digital radiographs of specific body parts. Advances in Digital Radiography: RSNA Categorical Course in Diagnostic Radiology Physics. Samei E and Flynn MJ eds (2003)
Seibert JA Digital radiographic image presentation: preprocessing methods. Advances in Digital Radiography: RSNA Categorical Course in Diagnostic Radiology Physics. Samei E and Flynn MJ eds (2003),63-70.
Chotas HG, Ravin CE. Digital radiography with photostimulable storage phosphor: control of detector latitude in chest imaging. Invest Radiol 27 (1992),
Huda W, Slone RM, Arreola M, Hoyle BA, Jing Z. Significance of exposure data recognizer modes in computed radiography. SPIE 2708 (1996),
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing

50. References (continued)
Freedman M, Pe E, Mun SK, Lo SCB, Nelson M. The potential for unnecessary patient exposure from the use of storage phosphor imaging systems. SPIE 1897(1993)
Gur D, Fuhman CR, Feist JH, Slifko R, Peace B. Natural migration to a higher dose in CR imaging. Proc Eighth European Congress of Radiology. Vienna Sep 12-17(1993)154.
Huda W, Slone RM, Belden CJ, Williams JL, Cumming WA, Palmer CK. Mottle on computed radiographs of the chest in pediatric patients. Radiology 199 (1996)
Radiation Protection in Digital Radiology L05 Digital Radiographic Image Processing