Quintiles Intelligent Imaging Clear Vision for the Healthcare Industry DICOM Grayscale Standard...

Quintiles Intelligent ImagingQuintiles Intelligent ImagingClear Vision for the Healthcare IndustryClear Vision for the Healthcare Industry

DICOM Grayscale Standard DICOM Grayscale Standard Display FunctionDisplay Function

David ClunieDavid Clunie

I2

OutlineOutline

• Inconsistent appearance of imagesInconsistent appearance of images– Why is it a problem ?Why is it a problem ?– What are the causes ?What are the causes ?

• Grayscale Standard Display FunctionGrayscale Standard Display Function– The DICOM solution to the problemThe DICOM solution to the problem– How it worksHow it works– How to implement itHow to implement it

Distributed Image ConsistencyDistributed Image Consistency

Digital Modality

Workstation

Laser Printer

Workstation

Identical perceived contrast

Distributed Image ConsistencyDistributed Image Consistency

Digital Modality

Workstation

Laser Printer

Workstation

Identical perceived contrastand color !!

What about color ?What about color ?

• Consistency is less of an issue:Consistency is less of an issue:– US/NM/PET pseudo-color; VL true color ?? US/NM/PET pseudo-color; VL true color ??

• Consistency is harder to achieveConsistency is harder to achieve– Not just colorimetry (i.e. not just CIELAB)Not just colorimetry (i.e. not just CIELAB)– Scene color vs. input color vs. output colorScene color vs. input color vs. output color– Gamut of devices much more variableGamut of devices much more variable– Greater influence of psychovisual effectsGreater influence of psychovisual effects

• Extensive standards efforts e.g. ICCExtensive standards efforts e.g. ICC

Problems of InconsistencyProblems of Inconsistency

• VOI (window center/width) chosen on VOI (window center/width) chosen on one device but appears different on one device but appears different on another deviceanother device

• Not all gray levels are rendered or are Not all gray levels are rendered or are perceivableperceivable

• Displayed images look different from Displayed images look different from printed imagesprinted images

• ……


mass visible mass invisible

•VOI chosen on one display device

•Rendered on another with different display

•Mass expected to be seen is no longer seen


0.5

1.5

1.0

3.0

•Not all display levelsare perceivable on alldevices


Digital Modality Laser Printer

•Printed images don’t looklike displayed images

Causes of InconsistencyCauses of Inconsistency

• Gamut of deviceGamut of device– Minimum/maximum luminance/densityMinimum/maximum luminance/density

• Characteristic curveCharacteristic curve– Mapping digital input to luminance/densityMapping digital input to luminance/density– ShapeShape– LinearityLinearity

• Ambient light or illuminationAmbient light or illumination

Causes of InconsistencyCauses of Inconsistency

1.0 .66

•Display devicesvary in the maximumluminance they canproduce

•Display CRT vs. film on a light box is an extreme example

Monitor Characteristic CurvesMonitor Characteristic Curves

Monitor Characteristic Curve

0.1

1

10

100

0 50 100 150 200 250 300Digital Driving Level

Ambient Light

MaximumLuminance

Gamma

Towards a Standard DisplayTowards a Standard Display

• Can’t use absolute luminance since Can’t use absolute luminance since display capabilities differentdisplay capabilities different

• Can’t use relative luminance since Can’t use relative luminance since shape of characteristic curves varyshape of characteristic curves vary

• Solution: exploit known characteristics Solution: exploit known characteristics of the contrast sensitivity of human of the contrast sensitivity of human visual system - contrast perception is visual system - contrast perception is different at different levels of luminancedifferent at different levels of luminance

Human Visual SystemHuman Visual System

• Model contrast sensitivityModel contrast sensitivity– assume a target similar to image featuresassume a target similar to image features– confirm model with measurementsconfirm model with measurements– Barten’s modelBarten’s model

• Grayscale Standard Display Function:Grayscale Standard Display Function:– Input: Just Noticeable Differences (JNDs)Input: Just Noticeable Differences (JNDs)– Output: absolute luminanceOutput: absolute luminance

Standard Display FunctionStandard Display Function

Grayscale Standard Display Function

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 200 400 600 800 1000 1200

JND Index



0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 200 400 600 800 1000 1200

JND Index

Monitors Film


.01

.1

1

10

100

1000

0 200 400 600 800 1000


JND Index


.01

.1

1

10

100

1000

0 200 400 600 800 1000


JND Index

Monitors

Film

Perceptual LinearizationPerceptual Linearization

• JND index is “perceptually linearized”:JND index is “perceptually linearized”:– same change in input is perceived by the same change in input is perceived by the

human observer as the same change in human observer as the same change in contrastcontrast

• Is only a means to achieve device Is only a means to achieve device independenceindependence

• Does not magically produce a “better” Does not magically produce a “better” imageimage


.01

.1

1

10

100

1000

0 200 400 600 800 1000


JND Index

Same number of Just Noticeable Difference == Same perceived contrast

Despite different changein absolute luminance


Modality

Display

Display Perception of ContrastBy Human Visual System

Ambient Light

Using the Standard FunctionUsing the Standard Function

• Maps JNDs to absolute luminanceMaps JNDs to absolute luminance• Determine range of displayDetermine range of display

– minimum to maximum luminanceminimum to maximum luminance– minimum to maximum JNDminimum to maximum JND

• Linearly map:Linearly map:– minimum input value to minimum JNDminimum input value to minimum JND– maximum input value to maximum JNDmaximum input value to maximum JND– input values are then called “P-Values”input values are then called “P-Values”

Monitor Characteristic CurveMonitor Characteristic Curve

Monitor Characteristic Curve

0.1

0

10

100

0 50 100 150 200 250 300Digital Driving Level

Ambient Light


.01

.1

1

10

100

1000

0 200 400 600 800 1000


JND Index

Monitor’s Capability

Jmax == P-Value of 2n-1

Jmin == P-Value of 0

Minimum Luminance+ Ambient Light

Maximum Luminance+ Ambient Light

Standardizing a DisplayStandardizing a Display

0.1

1

10

100

0 50 100 150 200 250

DDL or P-Values

Standard

Characteristic Curve


Mapping P-Values to Input of Characteristic Curve (DDL’s)

0

50

100

150

200

250

300

0 50 100 150 200 250 300

P-Values

DD

L


Standard Display Function

P-Values: 0 to 2n-1

StandardizedDisplay

Device Independent ContrastDevice Independent Contrast


P-Values: 0 to 2n-1


StandardizedDisplay B

StandardizedDisplay A

So what ?So what ?

• Device independent presentation of Device independent presentation of contrast can be achieved using the contrast can be achieved using the DICOM Grayscale Standard Display DICOM Grayscale Standard Display Function to standardize display and Function to standardize display and print systemsprint systems

• Therefore images can be made to Therefore images can be made to appear the same (or very similar) on appear the same (or very similar) on different devicesdifferent devices

So what ?So what ?

• Images can be made to appear not only Images can be made to appear not only similar, but similar, but the way they were intended the way they were intended to appearto appear, if images and VOI are , if images and VOI are targeted to a P-value output spacetargeted to a P-value output space

• New DICOM objects defined in P-New DICOM objects defined in P-valuesvalues

• Old DICOM objects and print use new Old DICOM objects and print use new services (Presentation State and LUT) services (Presentation State and LUT)

Not so hard …Not so hard …

• If you calibrate displays or printers at all, If you calibrate displays or printers at all, you can include the standard functionyou can include the standard function

• If you use any LUT at all, you can make If you use any LUT at all, you can make it model the display functionit model the display function

• If you ignore calibration and LUTs totally If you ignore calibration and LUTs totally (e.g. use window system defaults) the (e.g. use window system defaults) the results will be inconsistent, mediocre results will be inconsistent, mediocre and won’t use the full display rangeand won’t use the full display range

Quintiles Intelligent Imaging Clear Vision for the Healthcare Industry DICOM Grayscale Standard...

Documents

Transcript of Quintiles Intelligent Imaging Clear Vision for the Healthcare Industry DICOM Grayscale Standard...