CASCADED LINEAR-SYSTEMS ANALYSIS OF CMOS FLAT-PANEL DETECTORS FOR DIGITAL...

Biomedical Mechatronics Lab

CASCADED LINEAR-SYSTEMS ANALYSIS

OF CMOS FLAT-PANEL DETECTORS

FOR DIGITAL RADIOGRAPHY

Seung Man Yun, Min Kook Cho,Chang Hwy Lim, Ho Kyung Kim*

School of Mechanical Engineering, Pusan National University, Republic of Korea

Thorsten GraeveRad-icon Imaging Corp., Belick street, Santa Clara, CA 95045-2404 USA

Hyosung ChoDepartment of Radiological Science, Yonsei University, Republic of Korea

Jung-Min KimCollege of Health Science, Korea University, Republic of Korea

ANDE 2007

Cascaded Linear-systems Analysis of CMOS flat-panel detectors for digital radiography, ANDE 2007

Motivation

• For better design and usage of various radiographic modalities▫ Computed tomography (CT), digital radiography (DR)

Biomedical Mechatronics Lab 2

Miniatured cone-beam CT system using CMOS flat-panel detector

SO 11-III [2007-494] October 19, 17:00


CMOS flat panel detector

• Photodiode arrays manufactured by CMOS process with scintillator▫ Smaller image lag and larger fill factor than a conventional amorphous

photodiode array


Rad-icon RadEye1TM

Array format512 1024

pixels/1 detector

Pixel pitch 48 m

Field of view25 50

mm2/1 detector

Dynamic range 85 dB (>14 bits)

ADC bit-depth 12 bits

Pixel fill factor 0.87

Kodak Lanex Min-RTM

Composition Gd2O2S:Tb

Coverage 33.91 mg/cm2

Thickness 90 m

Density 3.77 g/cm3

RadEyeTM

Min-RTM


X-ray imaging system evaluation


DQE

NPS

Resolution

Contrast

Noise


Objectives

• Analyzing DQE of CMOS flat-panel detector for digital radiography▫ As a function of design parameters using cascaded model analysis

• Cascaded linear-systems theory-based modeling and simulation▫ Numerical modeling of NPS and DQE for x-ray imaging system evaluation

• Investigating a validity of the proposed cascaded model▫ Comparison with experimentally measured data for same condition

• Simulating DQE of the CMOS photodiode array with various design parameters▫ Photodiode quantum efficiency, fill factor, additive electronic noise



Evaluation procedure


)(NNPS

)(MTF)(DQE

0

2

fq

ff

2D FFT

1D FFT

• Slit images• Gain-offset corrections

• Synthesizing LSF`s

• HVL measurements• Spectral simulation• Tuning kVp• Estimating fluence

• Scaling for non-uniformity• Averaging• Extracting 1D profiles

• White images• Gain-offset corrections• Detrending• Conversion into relative noise• Windowing

f

f

f

mmAl

I/I 0

spect

ral density

MTF

1

DQ

E


Cascaded linear system analysis

• Cascade model describes the interacting of each process in detectors▫ The response of an system be linear and shift invariant (LSI), random noise

process be wide-sense stationary


Stage Description Symbol Process

Incident X-ray q Uniform distribution

Quantum detection g1=AQBinomial selection

Quantum amplification g2=AMBinomial selection

Quantum scattering TscnStochastic blurring

Quantum conversion g4=ADBinomial selection

Aperture integration TapertDeterministic blurring

Sampling III Deterministic process

Additive noise saddDeterministic process

1g

2g

scnT 3

gapert

T III addσ signaldigital q


• DQE of cascaded linear-systems of the CMOS flat-panel detectors


Fluence System gain

Noise power spectrum Additive electronic noise

System MTF

22

0

224

2222

σ)()(11

)()( )( )(DQE

addk

apertscn

M

DDMQ

apertscnDMQ

dd

kT

d

kT

I

AAAAAaq

TTAAAaq

ρρ

ρρρ


Monte Carlo simulation


Absorption Reflection

Refraction & transmission

Energy deposition& light conversionScintillator

Air gap = 1 m

Light photon detection plane

Spectral source sampling

MCNPXTM

DETECT2000TM

Polished surface

Grounded surface

Refractive index of phosphor = 2.6Absorption mfp. = 10 cmScattering mfp = 0.0017 cm

Pencil beam

Thin slap geometry: radius >> thickness

Subdividing the scintillatorinto thin sublayers to calculate the partial energy deposition and to estimate the escape probability of light photons with respect to the depth!!! Source

AED

OPD


Experimental setup

• RQA 5 experimental condition (IEC 62220-1)


Tube voltage 70 kVp (RQA5)

Tube current 0 ~ 125 mA

Exposure time at 1fps 550 ms

Source-to-detector distance 1000 mm

Added filter 21 mmAl

Added filter

Source-to-detector distance

X-raytube

CM

OS d

ete

ctor

surface

10 m wide slit camera(I.I.E. GmbH, Aachen, GER)

Slit camera

Ion chamber

Ion chamber(RAD-CHECK PLUS 06-526)


Cascaded model simulation parameter


Description Parameters

Pixel pitch d = 48 m

Pixel fill factor g = 87 %

pixel aperture a = 44.72 m

Incidence of X-ray q = 4.4×105 mm-2

Quantum absorption efficiency of screen AQ = 0.23

Average conversion efficiency of screen AM = 520

Quantum efficiency of photodiode AD = 0.55

Statistical swank factor I = 0.9

MTF of the scintillator Tscn( f ) = (1+1.0001f 2)-1

MTF due to the aperture integration Tapert( f ) = |sinc(af )|

Additive electronic noise sadd = 1100 e-


Model validation

• Compared with Kodak Lanex Min-RTM screen (RQA 5, @50mA)



Effect of a photodiode quantum efficiency



Effect of a fill factor



Effect of a additive electronic noise

• The DQE is vulnerable to the additive noise than the other parameters



Conclusion


• Cascaded linear-systems analysis of CMOS detector▫ Describes the signal and noise propagation

▫ Estimates its overall imaging performance

▫ Additive electronic noise is the most significant design parameter

• The developed model is a useful tool to design the CMOS flat-panel detector for digital radiography

• Based on this study, we can simulate a optimized design parameters of the CMOS photodiode array with scintillator

CASCADED LINEAR-SYSTEMS ANALYSIS OF CMOS FLAT-PANEL DETECTORS FOR DIGITAL...

Documents

Transcript of CASCADED LINEAR-SYSTEMS ANALYSIS OF CMOS FLAT-PANEL DETECTORS FOR DIGITAL...