uwe.ewert@bam.

de

Minimum Requirements for Digital Radiography

Equipment and Measurement Procedures by
Different Industries and Standard Organizations

Uwe Ewert and Uwe Zscherpel

BAM Federal Institute for Materials Research and Testing,
Berlin, Germany
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 1
 Introduction

- Film Radiography has been developed for quality assurance in industry

over decades.
- It is basis for contractual agreements of Supplier and Purchaser.
- Digital Radiography shall achieve the same image quality or better, but
with higher efficiency.
- Selected image quality requirements of different standards as ASTM,
ASME and ISO are compared.
- The essential parameters for optimization and prediction of IQI visibility
are discussed.

- Requirements for equipment selection are derived for CR and DDAs.

- Newly developed classification procedures for RT-D equipment are
introduced.

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 2
 Basic Requirements in Radiography
Requirement Film Digital
ISO, CEN, ASTM,
Standards ISO, CEN ASTM, ASME, MAI
ASME
DDAs: Exceed
Exceed minimum minimum required
Exceed minimum required SNRN CNR
Exposure
optical density (Normalized Signal
to Noise Ratio) CR: expose to EPS
plateau (MAI)
Use detectors which
achieve the required
Use required film Do not exceed
Detector SNRN and
system class or image
requirements do not exceed the
better unsharpness limits
detector
unsharpness limit
Do not exceed the Do not exceed the Do not exceed the
Maximum image
geometrical image image
unsharpness
unsharpness limits unsharpness limits unsharpness limits
Achieve IQI contrast
Achieve required IQI Achieve required IQI
Image quality sensitivity + duplex
contrast sensitivity contrast sensitivity
wire resolution
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 3
 Key Technologies for Film Replacement
• Computed Radiography (CR) with storage
phosphor imaging plates

• Digital Detector Arrays (DDA)

CR DDA

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 4
 Image Quality in Radiography
Intensity
- Influence of SNR and CNR -

Intensity
Contrast Contrast

Signal Signal
(base material) (base material)

Length Length

Notch visible! Notch not visible!

Notch is visible
Contrast/Noise is high Contrast/Noise is low in the profile if
Signal/Noise is high Signal/Noise is low C > 2.5  Noise
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 5
 Noise Sources in Radiographic Images

Typical noise sources in digital radiography:

1. EXPOSURE CONDITIONS: Photon noise, depending on
exposure dose (e.g. mAs or GBqmin). This is the main factor!
SNR increases with higher exposure dose.
2. Limitation for the maximum achievable SNR:
1. DETECTOR: Structural noise of DDAs and Imaging Plates
also called fixed pattern noise (due to variations in pixel to
pixel response and inhomogeneities in the phosphor layer).
2. OBJECT:
1. Crystalline structure of material (e.g. nickel based steel,
mottling)
2. Surface roughness of test object

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 6
 Measurement of Contrast to Noise Ratio in DDA Practice
by ASTM E 2698
• CNR shall be measured in the
Noise I=contrast 4T hole for proof of image quality.
• A minimum CNR of 2.5 is
required by ASTM E 2698.
• This value needs to be revised!

ASTM E 1025

I CNR = 6.7
C = I = 473
Noise = 71
SNR = 155
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 7
 Human Observer Model: Perception Threshold PT
50 µm pixel size ASTM E 2698
requires CNR > 2.5
Diameter Noise = 1000
Signal = 30000
Needs revision!
0.12 0.25 0.5 1 2 mm
CNR = 2.5  Large area flaws are better
d - diameter visible than small ones at
of hole same Contrast-to-Noise
Ratio
CNR = 1.25
 Each row has indications
with same CNR = Cdepth/
CNR = 0.625
Flat bottom holes of different depth and diameter  Each column has holes
SRb image = 50 µm with same diameter

d PTconst - constant human perception threshold  10

PTconst   CNR d – diameter of just visible hole
SRb CNR - contrast to noise ratio
SRb – basic spatial resolution (effective pixel size)
Rose approach, 1946
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 8
 System Selection for ASTM Standard Practice
 ASTM Draft, “Practice for the Use of Digital Detector Arrays and Computed
Radiology for Aerospace Casting Inspections” requires
 a minimum basic spatial resolution of the detector (SRbdetector) if images
have been taken without magnification or
 a minimum basic spatial resolution in the magnified image SRbimage).

Effective pixel size of digital image ( SRimage )

b

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 9 9
 Verification of CNRmin for ASTM Standard Practice
 ASTM Draft, “Practice for the Use of Digital Detector Arrays and Computed
Radiology for Aerospace Casting Inspections” and E 2698 “Standard
Practice for Radiological Examination Using Digital Detector Arrays” require
for digital images with DDAs in the 2T IQI hole:

CNRmin  2.5
 The ASTM E 2698 value (CNR = 2.5) is optimized for testing of thin objects
with thickness < ½” and typical DDAs in the range of 0.1 mm < SRb < 130
µm.
 Generally, CNRmin for all digital images can be described more accurate by
the equation:
10  SRbimage
CNRmin 
diameterIQI  hole
Based on “old” Rose equation as shown before for PTconstant

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 10
 Contrast Sensitivity as Required by
Different Standards
 Almost all standards require image quality
indicators on each production radiograph.
 The operator has to evaluate if the required image
quality has been achieved.
 He decides about acceptance or rejection of the
production radiographs.
 Different international standards require different
contrast sensitivities, which yields different
inspection quality.
 The thickness sensitivity improves with material
thickness.
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 11
 Basic Requirements for IQI Visibility of International
Standards in Comparison
ISO 19232-3
Step Holes, ISO 17636-2, 2013
Plate Holes, ASME BPVC SC V Ar. 2
Plate Holes, ASTM E 1742, E 2104
ASME BPVC SC V, Ar. 2, and
ISO 17636-class A are about
equivalent

ASTM E1742, E 2104,

ISO 17636-class B are about
equivalent below t=12 mm (1/2“)

ASTM E 1025

ISO 19232-2

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 12
 Additional Requirements for Unsharpness
and Basic Spatial Resolution

 In all film radiography standards the permitted geometric

unsharpness (Ug) is limited and the film unsharpness is neglected.

 In digital radiography the detector unsharpness (Udetector) contributes

significantly to the image unsharpness (Uim) or total unsharpness
(UT).

 Therefore, the permitted detector unsharpness (Udetector) is limited in

different standard systems, which is relevant for detector selection.

 The detector unsharpness shall be always smaller than the permitted

geometric unsharpness (Udetector < Ug).

 The basic spatial resolution is defined as ½ unsharpness in digital

radiography (SRb = ½ U).

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 13
 Measurement of Basic Spatial Resolution

Duplex wire IQI ASTM E 2002

ISO 19232-5 Two new
ASTM E 2002 wire pairs

Duplex Wire IQI, 90 kV,

1m SDD, small focus

• The detector unsharpness

udetector shall be controlled
by reference exposures with
the duplex wire IQI.
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 14
 Standard ISO 17636-2 (2013): Table B.13, B.14
Maximum detector or image unsharpness (2SRb)

DDA

CR
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 15
 Unsharpness Requirements of Different International
Standards – No International Harmonization -
Unsharpness requirements

0,08
E 1030, 1032
0,07
Unsharpness in inch

0,06
E 1030 (2004)
0,05

0,04 MAI
E 2104
0,03
E 1742 ISO 17636 class A
0,02
ISO 17636 class B
0,01

0,00
0,0 1,0 2,0 3,0 4,0 5,0 6,0
Thickness in inch Co-60, High Energy

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 16
 Unsharpness Requirements in Different Int. Standards

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 17
 Qualification of CR-Systems
EPS= equivalent penetrameter sensitivity (see ASTM E 746, E 747, E 1025) - The EPS Concept -
EPS – Procedure is proposed accepted for CR qualification in
ASTM E 20445/6 draft and the draft on “Practice for the Use of … Computed
Radiology for Aerospace Casting Inspections” (USA: MAI – group)

- The EPS (equivalent penetrameter sensitivity) measurement is based on

ASTM E 746
 A smooth ¾ inch (19 mm) steel plate with a set of
plate holes is radiographed at 200 kV in ≥ 1 m
distance
 The exposure is performed with different mAs
settings
 A graph is generated, see next pages
 The calculation of just visible hole diameter is
given by:
SRbimage
d visible  PT 
 eff  SNR ASTM E 746
PT depends slightly on operator and viewing conditions
New formula µeff for 200 kV and 19 mm Fe is about 0,05 mm-1
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 18
 Example: EPS Test with HD CR Scanner at 20 µm Pixel Size

Do you see the holes?

¾” (19 mm) steel plate, 200 kV

1 mm steel step for
CNR  0.8
measurement of µeff

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 19
 Visibility of EPS holes (E 746)

Do you see the holes?

CNR  0.8
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 20
 New Formula for Conversion of SNRN Measurements to
EPS Values and Working Range for CR (ASTM Draft E 2033)

 PT’ is about 2▪100 for visibility of the 2 T hole of IQIs corr. to ASTM E 1025
 EPS by ASTM E 746 with 200 kV, t = 19 mm Fe plate and µeff = 0.05 mm-1
EPS method for determination of performance levels
Measured EPS,IP 1, Scanner 2 Best fit Level I level II
2,6

PT ' SRbimage
EPS 
2,4

µeff  SNR
140
2,2 ttestplate
EPS

2
350
1,8 Level II: PV 140 – 4095
1,6
aEPS = 1.16
1,4
Level I: PV 350 – 4095
1,2

1 90% PVmax
0 500 1000 1500 2000 2500 3000 3500 4000 450

Plateau range Pixel value 

0 23 46 69 91 114 137 160 183
206
Dose / mGy
EPS= equivalent penetrameter sensitivity (see ASTM E 746, E 747, E 1025)
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 21
 Old CR Classification Scheme

System class Minimum normalised

CEN SNR Classification scheme of
ASTM; e 2446, EN
IP 1/Y 130
14784-1 and ISO 16371-1
IP 2/Y 117
by system classes.
IP 3/Y 78
IP 4/Y 65 The CR systems are
IP 5/Y 52 classified by the
IP 6/Y 43 maximum achievable
Note 1: The normalized SNR values of Table 1 are SNRN (IP 1 – 6) and the
similar to those of EN 584-1. They are calculated by
SNR= log(e) (Gradient/Granularity) of Table 1 in EN
basic spatial detector
584-1. The measured SNR values are calculated from resolution SRbdetector (Y).
linearised signal data.

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 22
 New CR Classification Scheme
New classification by performance levels as given in the
draft revision of ASTM E 2446 (2014).
Additionally, a specified EPS performance is required.

CR System Minimum Maximum Maximum

Classification SNRN iSRbdetector achieved EPS
(normalized value [µm] by E 746 [%]
to SRb=88.6
µm)
CR Special 200 50 1.00
CR Level I 100 100 1.41
CR Level II 70 160 1.66
CR Level III 50 200 1.92
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 23
Qualification “Spidernet” Graph and Classification Statement

Qualification of CR‐System XY 
Interpolated SRbdetector for
E 2795, ISO 17636-2 users
SRbdetector = 145µm
14
Achievable SNRN for 12 Achievable CS for
ISO 17636-2 user 10 E 2597 user
aSNRN = 88 8 Csa = 0,8
6
4
2
0 Performance level II
Speed
as defined for film
ISO Speed@SNRN130 = 
aEPS@mag 1 = 1,65
400
Achievable EPS
Specific for CR

Efficiency@1mGy = 
115
Efficiency as defined for DDAs in E 2597

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 24
 New Qualification Procedures
Required Test for ASTM E 2445/2446 Required Result
Geometric Distortion Fail if distortion > 2%
Laser Jitter Not permitted,
Laser Beam Scan Line Integrity Not permitted
Scan column dropout Not permitted
Scanner Slippage Not permitted
Imaging plate Artifacts Not permitted
Erasure Fail if > 2%
Shading or banding Fail if more than ±10%

Test results shall be reported, Result to report

classification is possible in case of exceeding the limits
PMT Non-linearity Report if > 2%
Burn-In Report if > 2%
Spatial Linearity < 2%
Imaging plate response variation Report if > ±10%

Optional test on request Result to report

Imaging Plate Fading (no test object required), optional test Report fading in %,

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 25
 Requirements for Digital Detector Arrays

Qualification by ASTM E 2597M

DDA

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 26
 BAM 5, 8mm steel

FujiFilm IX25
SNRN ~ 265

DDA Technology
provides better image
Best (slowest) NDT film quality than film with a
special calibration
procedure!

Images high pass

filtered for better
presentation

PerkinElmer 1620
SNRN ~ 1500
Magn. = 3.5
DDA
8.3 exposure
Radiological
Methods
ECNDT
Praha, Oct. 2014
Requirements for RT-D
Ewert and Zscherpel 27
 Qualification of Digital Detector Arrays
- Management of Underperforming Pixels -
A bad pixel can be corrected if it has at least
5 good neighbors
ASTM E 2597M
Definition and Test of Bad
Pixels:
6.2.1.1 Dead Pixel
6.2.1.2 Over Responding Pixel
6.2.1.3 Under Responding Pixel
6.2.1.4 Noisy Pixel
6.2.1.5 Non-Uniform Pixel
6.2.1.6 Persistence/Lag Pixel
6.2.1.7 Bad Neighborhood Pixel

6.2.2 Types or Groups of Bad

Pixels:
6.2.2.1 Single Bad Pixel
6.2.2.2 Cluster of Bad Pixels
6.2.2.3 A cluster without any CKP is
well correctable.
6.2.2.4 A cluster with CKP is labeled a
relevant cluster.
6.2.2.5 A single bad line segment is a
special irrelevant cluster.
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 28
 Qualification of Digital Detector Arrays
- Pixel Coverage Recommendations -

ASTM E 2736

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 29
 Qualification of Digital Detector Arrays
- Management of Underperforming Pixels -

ASTM E 2736

Compensation
Principles

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 30
 Compensation Principle (II) of ISO 17636-2

Compensation of high detector unsharpness by increased SNR

• Unsharp digital systems may be applied for NDT if they enable
to compensate the missing sharpness by increased SNR.
• That means, achieves a digital system not the required
visibility of the separated duplex wires, it can be used for
NDT, if one or two single wires more than required (see
tables B.1 –B.12 of ISO/DIS 17636-2) can be seen clearly in
the digital image for one or two missing duplex wire pairs.
Compensation of 3 wires vs. wire pairs requires agreement
of contracting parties.
• Compensation principle (II):
• High detector unsharpness can be compensated by
increased SNR 31

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 31
Example : Compensation Principle (II) of ISO 17636-2

Interesting for detectors with higher unsharpness

Compensate missing spatial resolution by increased single

wire sensitivity:
• A lower spatial resolution i.e. a lower double wire score (D)
may be compensated by a higher single wire sensitivity i.e.
higher single wire score (W).
• Max. two (or three) single/double wire scores may be exchanged.

Not Required: OK: OK OK

OK by agreement
Duplex wire score D13 D12 D11 D10 D9
Single wire score W13 W14 W15 W16 W17

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 32
 Qualification of Digital Detector Arrays
- Efficiency Measurement -

120 kV,
3 mm Cu

120 kV, 40 mm Al
160 kV

50 kV
90 kV

Qualification by
ASTM E 2597M
220 kV, 40 mm Al

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 33
 Qualification of Digital Detector Arrays
- Image Lag -
Qualification by ASTM E 2597M

Important for CT and fast

measurements as e.g.
automated defect
recognition (ADR).

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 34
 Qualification of Digital Detector Arrays
- Contrast Sensitivity by CNR-Step Wedge Measurement -
Qualification by ASTM E 2597M

Step wedge with notch

Measure
contrast and
noise per step

Contrast sensitivity (CA) at 5% notch in a step wedge by 1/CNR

Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 35
 Qualification of Digital Detector Arrays
- Contrast Sensitivity by CNR-Step Wedge Measurement -
Qualification by ASTM E 2597M

Working range (4s)

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 36
 Qualification of Digital Detector Arrays
- SMTR by SNR-Step Wedge Measurement -
Qualification by ASTM E 2597M

SNR = 250

SNR = 130

Specific Material Thickness Range by SNR limits

Qualification by ASTM E 2597M

Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 37
 Qualification of Digital Detector Arrays
DDAs are qualified by different tests of ASTM E 2597.
Five relevant parameters shall be provided:
 Basic spatial resolution (SRb)
 Specific Material Thickness Range
 Efficiency Qualification by ASTM E 2597M
 Image lag
 Contrast sensitivity

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 3838
 Conclusions
 Replacement of film radiography by digital techniques in NDT requires careful selection of
suitable digital detectors.
 Standards on digital radiography were published and revised since 2005 in ISO, CEN, ASTM
and ASME, defining requirements for image quality and detector selection.
 International standards define different requirements for image quality and detector selection.
 The image quality depends on the essential parameters: specific contrast µeff, SNR and basic
spatial resolution SRb.
 CR is accepted as film replacement technology:
 The prove of image quality requires SNRN measurements and IQI visibility (ISO 17636-
2).
 CR classification and qualification is under major revision.
 Classification will consider SNRN, SRbdetector and EPS
 DDAs achieve a significantly better contrast sensitivity with suitable detector calibration than
film radiography.
 The prove of image quality requires CNR measurements and IQI visibility (ASTM practice
E 2798).
 DDAs are qualified by different but similar procedures than CR systems, because the DDA
detector calibration and image integration influences seriously the qualification, and DDAs
provide currently better image quality than CR.
Requirements for RT-D
8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 39
 End
uwe.ewert@bam.de

Requirements for RT-D

8.3 Radiological
ECNDT
Methods Praha, Oct. 2014 Ewert and Zscherpel 40