You are on page 1of 54

DR system:

How to find the best hardware and software to


achieve minimum dose and optimal image quality
Radiographer, M.Sc. Health: Helle Precht & PhD: Oke Gerke
Denmark

RC 1414 - Paediatric imaging

Agenda
Canons indirect DR system:

DR hardware

Project focusing on scintilator sensitivity in paediatrics

Possibilities in software processing (Multi Frequency Processing)

Optimization

Project focusing on possibilities to grade dose and image quality in


relation to the type of the paediatric examination

Educating Tomorrows Professionals www.ucl.dk /conradint

Cooperation between:

Educating Tomorrows Professionals www.ucl.dk /conradint

DR hardware

DR Hardware

Connection to generator

Computer capacity

Connection to RIS/PACS systems

Touch screen

Diagnostic monitors

Detector

Educating Tomorrows Professionals www.ucl.dk /conradint

DR hardware

DR detector

Stationary/portable ~ Wireless

DQE and MTF

Pixel size and fill factor

Fig.: DR detector (Bushong, 2004)

Fill factor = Light sensitive area/detector area

Scintilator line spread function

GOS

CsI

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Project focusing on
scintilator and scoliosis
Background

Scoliosis pathology
Human radiation response in relation to patient age
Tissue weighting factor and contact shield
ALARA
Technique: high kV, airgap and stitching
GOS and CsI scintilator

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Hypothesis
A Canon detector with CSI scintilator will
produce acceptable image quality at a
scoliosis examination at lower dose than a
Canon detector with GOS scintilator

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Method
Theory supported by published articles, books and
information by Canon
Quantitative experimental design
Canons CXDI 50G and 50C detector
Human phantom (audit)
Dosimeter (DAP and ESD - Unfors)
Monte Carlo dose calculations
Statistics

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Results
200

350

180
300

160
250

120
200

REX

DAP mGy cm^2

140

100

150

80
60

100

40
50

20
0
0,5

0,6

0,8

1,2

0
1,6

DAP 1 GOS

2
0,5

2,5
3,2
0,6
0,8
mAs

DAP 2 GOS

4
1

DAP 1 CsI

6,3
1,2

10
1,6

12,5
2

16
2,5

20
3,2

25
4

6,3

10

12,5

16

20

25

mAs

DAP 2 CsI
REX 1 GOS

REX 2 GOS

REX 1 CsI

REX 2 CsI

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Analyzing image quality


Participants: three radiologists and three reporting radiographers
Every image was scored according to:
No.
Def.:

Too low SNR

Acceptable SNR

High SNR

Reduced spatial

Acceptable spatial

High spatial resolution

resolution

resolution

Image criteria not

Image criteria barely

Image criteria above

met

met

requirements

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Results

Radiographers

2
1

CsI

20

16

CsI

GOS

25

12,5

10

CsI
CsI

mAs

GOS

0,5
0,6
0,8
1
1,2
1,6
2
2,5
3,2
4
6,3
10
12,5
16
20
25

0,5
0,6
0,8
1
Number of score
1,2
1,6
2
2,5
3,2
4
6,3

Number of score

Radiologists

mAs

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Technical measurements of
sensitivity at different kV levels

CsI, CXDI 50C


GOS, CXDI 50G

Fig.: CsI and GOS scintilators signal reinforcement at different kV values.


Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Conclusion
The REX value can be used as an objective indicator of image quality based on the
indication for the examination. Dose and scintilator amplification degree affect REX
value.
Based on the experiments the hypothesis is confirmed:
The CsI detector can at 2 mAs produce an acceptable image quality, where GOS does
not produce comparable image quality until 6,3 mAs. This confirms the theory about
the CsI detectors DQE and higher REX value compared to the GOS detector at all mAs
levels.

Educating Tomorrows Professionals www.ucl.dk /conradint

Scintilator project

Perspectives
Other DR products?
Technical phantom for more objective results

Use of other modalities - CT, MRI or UL?


Software optimization

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Pre - and post processing

Fig.: Process in production of image data (Canon Inc., 2008a).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Automatic histogram adaptation

Fig.: Exposure recognition (Canon Inc., 2008b)

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

LUT curves in Canons DR system

Bone#1

Bone#2

Chest

Standard

Inv Linear

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Index value: REX - ROI

ROI

Fig.: Basis for REX calculation (Canon Inc., 2001).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a human phantom

REX: 257

REX: 641

REX: 655

REX: 4747

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Guide for software optimization


Turn off all the functions of MLT(S)
Set LUT
Adjust ROI Contrast

Max. density region is dark

Min. density region is bright

Graininess needs to be reduced

Sharpness is not enough

Adjust Dynamic Range Dark Region

Adjust Dynamic Range Bright Region

Adjust Effect of Noise Reduction

Adjust Frequency Band

Adjust Effect of Edge Enhancement

Fig.: MLT(S) flow chart (Canon Inc., 2008a; Canon Inc., 2008b)
Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Dynamic range

Bright region

Dark region

Fig.: Technical illustration of MLT(S) compression (Canon Inc., 2008a).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a human phantom

Dynamic range, Dark region

20

Dynamic range, Bright region

20

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Contrast local and global

Fig.: Local and global contrast (Canon Inc., 2008a).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a CD Rad phantom

30
Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a human phantom

30
Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Frequencies

Fig.: The building and function of the laplacian pyramid (Vuylsteke, Schoeters, 1999).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Edge enhancement

Fig.: Unsharp masking (Gonzales, Woods, 2008).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a CD Rad phantom

Effect: 1

Frequency: 1

Effect: 20

Frequency: 7
Frequency: 4
Effect: 20

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a human phantom

Frequency: 1

Frequency: 7
Effect: 20

Effect: 1

Effect: 20
Frequency: 4

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Noise reduction

Fig.: The principle behind low pass filtration (Gonzales, Woods, 2008)

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a CD Rad phantom

10

Educating Tomorrows Professionals www.ucl.dk /conradint

Software processing

Example using a human phantom

10
Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Project focusing on
software optimization
Background

Survey of optimization level with Canons European Application Group

New MLT(S) software new possibilities within Radiography?

Lack of research in Paediatric Radiography within the software optimization

Are we in accordance with national and international standards?

Do we use the full potential of DR systems?

Radiographers job description

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Survey (Questionaire)

Implementation - dose
6
5
4
3
2
1
0

Cause of missing optimization


Number of answers

6
5
4
3
2
1
0

Number of answers

Number of answers

Implementation - image quality

8
7
6
5
4
3
2
1
0

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Hypothesis
Hyp.1:
With Canon's new MLT(S) software one can maintain
optimal image quality at lower mAs in paediatric
examinations of the femur.
Hyp.2:
If the pathological focus at a femur examination is
changed from primary to follow-up examination of a
fracture, it is possible to reduce mAs more than the
achieved mAs value from hyp. 1 using MLT(S).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Optimization
Current practice

Adjust practice
and formulate
possible new
criteria for
good practice

Compare practice
with criteria for
good practice

Point out deviation

Fig.: Quality development as a dynamic process (Kjrgaard, 2001)


Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Optimization in Radiography
1.

Set of standards, always based on an


anatomical background

2.

Make sure that these live up to image quality


(diagnosis) and dose demand (reference dose)

3.

Optimize low performing practices

4.

Set/develop new standards

5.

Repeat
(European Commission, 1996a; Bth et al., 2005)

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Which examination to start optimizing?


The following examinations should be given priority in
the optimization of paediatric examinations:

Acquisition leading to repeated radiation


Acquisition that gives high radiation dose
Acquisition involving radiation sensitive area
(ICRP, 2006; ICRP - annals of the ICRP, 2004).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

As Low As Reasonable Achievable


(ALARA)
diagnostic
information
Dose saturation

dose
Threshold value for diagnostic information
Fig.: Dose draft on diagnostic information (Norrman, 2007)

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Method
Theory supported by published articles,
books and information by Canon

Quantitative experimental design


Technical phantom (CD Rad)
Human phantom (VGA analysis)
mAs and software settings are variable
Monte Carlo dose calculations
Statistics

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Phantoms

Technical CD Rad phantom with water absorption

Human lamb phantom

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Technical parameters used


in the experiments
Fixed parameters:
60 kV
Total filtration: 4,2 mm Al
SID: 100 cm
LUT: Bone#1
Collimation: 42x42 cm and 26x13 cm
Variable parameters:

16-0,5 mAs
Software parameters

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Software settings
Software processing
Contrast
Dynamic range, Dark Region

Settings
16
10

23
13

29
16

20

Noise reduction

10

Edge enhangement, frequency band

Edge enhancement, effect

10

Table: Applied MLT(S) parameters in CD Rad tests

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Analysis of CD Rad images

Fig.: CD Rad analyser (Artinis, 2006)

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Analysis of human phantom images:


1st hypothesis
1

VGA

Visualization of age appropriate trochanter, femoral head, medial and


lateral condyle and femur bone

Visualization of periarticular soft tissue level

Sharpness of the demarcation between cancellous bone and compact


bone

Sharpness of trabecular

Table: Image criteria on femur images (Bontrager, 2002; European Commission, 1996b)

-2

Clearly worse than the reference image

-1

A little worse than the reference image

Comparable with the reference image

+1

A little better than the reference image

+2

Clearly better than the reference image

Table: Relative VGA scale for scoring image quality (Almn, et al, 2000).
Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Analysis of human phantom images:


2nd hypothesis

VGA

Visualization of the fracture bone ends of the


femur and their position

Table: Image criteria for control exposure of femur AP.

Not visible

Poorly reproduced

Well reproduced

Very well reproduced

Table: Absolute VGA scale for scoring image quality (Almn, et al, 2000).

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Results

S-4

S-10
2 mAs
Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Results
Radiologist

Not visible

Poorly
represented

Well
represented

Very well
represented

56 (46.67%) 56 (46.67%)

14 (11.67%)

78 (65%)

28 (23.33%)

1 (0.83%)

33 (27.5%)

86 (71.67%)

8 (6.66%)

Table: Frequency table on the radiologists score, number (%), of the image criteria
for each of the four scoring possibilities within the hypothesis.

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Statistical results
Significant factors by 1th hypothesis:
mAs
dynamic range, dark region
frequency band

Significant factors by 2nd hypothesis:


mAs
dynamic range, dark region
frequency band
edge enhancement effect

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Bias
1. During the CD Rad tests the use of two images at each adjustment of
MLT(S) parameters and dose was an absolute minimum, the
recommendation is six identical images.
2. Calculation of applied water phantom as an absorber to the CD Rad
phantom.
3. Use of a lamb phantom; the difference to human anatomy is natural.
4. Size and absorption of the human phantom was larger than femur of a
five-year old child.
5. Manually placing ROI and its influence on the REX value.

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Conclusion
Distinction between optimal and diagnostic image quality.
Based on the experiments both hypothesis is confirmed:
Optimal image quality is obtained at a dose reduction of 70 % from 16 to 5 mAs with
MLT(S) optimized images. Specifically optimized images are approved at 2 mAs, but
the radiologists VGA scores are worse than the reference image (diagnostic image
quality). This reduction consists of 97 %.

In follow up exposures of femur fracture all the radiologists approved optimized


images at 0,5 mAs corresponding to a dose reduction of 92 %.
Because of the factual bias of the project it might generally be possible to reduce dose
even further, as the lamb phantom absorbs more radiation than a five year old child.

Educating Tomorrows Professionals www.ucl.dk /conradint

Software project

Perspectives
New version of the MLT(S) software.
The complexity of software optimization demonstrates the necessity of more educated
radiographers with a view to handle development and implementation of such practices.

Future software could incorporate processing combinations designed for representing a given
pathology optimally with the lowest possible dose.
In the future examine possibilities of the software in several organs, pathologies and patient
groups a manual on software optimization will be developed as well as a database on applied
radiographic techniques and software settings for all Europe.

In order to disseminate the achieved knowledge two articles will be written for publication in
Paediatric Radiology.

Educating Tomorrows Professionals www.ucl.dk /conradint

References

Almn, A., Tingberg, A., Mattsson, S. et al. (2000); The influence of different technique factors on
image quality of lumbar spine radiographs as evaluated by established CEC image criteria, The British
Journal of Radiology, vol. 73, pp. 1192-99.
Artinis (2006); Manual Contrast-Detail Phantom, Artinis CD Rad type 2.0.
Bth, M., Hkansson, M. et al. (2005); A conceptual optimisation strategy for radiography in a digital
environment, Radioation Protection Dosimetry, vol. 114 pp. 230-35.
Bontrager, K.L. (2002); Textbook of Radiographic Positioning and Related Anatomi, 4 th edn, Bontrager
Publising, Phoenix.
Canon Inc.(2001); X-ray Digital Camera CXDI Series, Technical guide Image Processing, Japan.
Canon Inc. (2008a); CXDI Image Processing Software MLT(S) Users Manual, Japan.
Canon Inc. (2008b); Multiobjective Frequency Processing Function manual MLT(S) Edition, Japan.
European Comission (1996a); European guidelines on quality criteria for diagnostic radiographic
images, Luxemburg.
European Commission (1996b); European guidelines on quality criteria for diagnostic radiographic
images in paediatrics, Luxemburg.
Gonzales, R.C. & Woods, R.E. (2008); Digital Image Processing, 3 rd edn, Pearson, Prentice Hall.
ICRP (2006); Recommendations of the International Comission on Radiological Protection.
ICRP annals of the ICRP (2004); Guest Editiorial Managing patient dose in digital radiology, vol.
34, pp. 1-73.
Kjrgaard, J. (2001); Kvalitetsudvikling i sundhedsvsenet, 1 st edn, 3 rd oplag, Munksgaard, DK.
Norrman, E. (2007); Optimisation of radiographic imaging by means of factorial experiments
Doctoral Dissertation, rebro studies in Phisics 3, rebro University, Sweden.
Vallgrda, S. & Koch, L. (2007); Forskningsmetoder i folkesundhedsvidenskab, 3 rd edn, Munksgaard,
Copenhagen.
Vuylsteke, P. & Schoeters, E. (1999); Image Processing in Computer Radiography. Vol. 16 pp 87-101.
Educating Tomorrows Professionals www.ucl.dk /conradint

Questions?

Denmark
Thanks for your attention
E-mail: hepr@ucl.dk
Educating Tomorrows Professionals www.ucl.dk /conradint

You might also like