CCD Digital Radiographic Detectors

Michael Yester, Ph.D.

University of Alabama at
Birmingham
 Outline

• Introduction
• CCD Fundamentals
• Characteristics related to imaging
• Implementations in Radiology
• Advantages and Disadvantages
• CMOS
• Conclusion
 Introduction

• CCDs are quite widespread

– used in digital cameras
– videocameras
– Astronomy
– Photomicrography
• Some of original digital devices used in Radiology
• Relatively simple
• New offerings of CCD technology in Radiology
 Introduction

• CCDs are limited in size

• CCDs are indirect type of digital receptor
• Have optical considerations in addition to usual
considerations
• Related to disadvantage
 CCD Fundamentals

• Device is a silicon chip

• Photosensitive layer
• Embedded electronics
 CCD Fundamentals

• Polysilicon layer contains the gates (in essence the pixels)

• Silicon dioxide acts as an insulator
• Bulk silicon contains depletion area and charge storage area
 CCD Fundamentals -- Sequence of events

• Light strikes device

• Electron-hole pairs formed
• Electrons constrained to an
area by electrostatic forces

• Each pixel contains 3

electrodes as shown
• Charge readout in “bucket
brigade” fashion
 CCD Architectures

• Full-frame -- frame dumped out as a unit

– Simpler and used in general radiological applications
– Entire surface area of chip utilized so fill factor = 100%
– An entire scene is captured (e.g. x-ray exposure)
• Interline -- readout and photo-detection in one area
– Readout is shielded and charge is transferred to the
readout area at a given frame rate (shutter control)
– High frame rates but fill factor is significantly reduced
 CCD Fundamentals -- Readout
Full Frame

• Move charge down column

using voltage sign changes
• Charge eventually reaches
readout row
• Charge in readout row (serial
register) is sequentially moved
out to sense amplifier and
digitized
• All of above takes place under
control of clock signals
 CCD and blooming

• Each cell has a given well depth for the storage of the charge
• If the charge exceeds the well depth, electrons will spill into
adjoining cells
• Design chip to accommodate expected well depth
• In some applications overflow drains are incorporated into the
chip
 Vertical Overflow Drain

• Vertical -- set barrier with

height less than surrounding
• adds to complexity and may
reduce overall dynamic range
• (reduces original barrier
height)
 Lateral Overflow drain

• Lateral -- uses part of the chip

to store the excess
• reduces fill-factor

Adapted from Kodak CCD Primer - #KCP-001

 CCD Characteristics and Imaging

• CCD chip size is limited to a maximum of 5 x 5 cm

• Cost is dominant issue
– Number of acceptable chips is low
– Large chip quite expensive, acceptability rates low
– Cheaper to make a large number of smaller chips
• Demagnification needed in most applications
– Match light emission source and CCD size with optics
(lenses or fiber optics)
• Amount of light incident and subsequently absorbed is
very important
 Important components and characteristics

• Scintillator -- amount of light input

• Optics -- capture of light and transmission to chip
• CCD quantum efficiency -- light absorption, spectral
sensitivity
• Noise
 Scintillator

• Absorption of x-rays and light conversion

• Spectrum of emitted light

• Characteristics are not unique to CCD

– Spectrum is important to CCD
• Common to any indirect digital system
• CsI is common
– “needle” crystalline structure advantageous
 Optical considerations

• Light transport efficiency is critical issue

• Transmission of light through medium (will depend on
spectrum)
• Focusing media
– lenses
– fiber optics
• Different issues for each
 Lenses

• Efficiency of transmission
– ~ 0.7 to 0.8 for CsI spectrum (550 nm)
• Focal length
• Demagnification factor
• Overall collection efficiency
 Collection efficiency for lenses

TL
ηL =
1+ 4 • f # • (1+ m)
2 2

• Assuming Lambertian source of light

TL = transmission factor of lens
f# = f-number of lens
m = demagnification factor
• Note f# and m appear as squared terms
Example
f# = 1.2, TL = 0.8, and m=2
ηL = 1.5%
 Other issues with lenses

• Geometric distortion
• Vignetting
• Veiling glare (light scatter)

• Main concern is coupling efficiency and creation of a

secondary quantum sink
• Note: it is possible to avoid the above but much care
and effort involved, especially at low exposure
 Fiber optics

• Want to have 100% internal

reflection
• For a bundle have to consider
ratio of length-to-diameter
(normally a high ratio)
• Any loss will be magnified
 Collection efficiency -- fiber optic
2

ηTFO =
1
2

•
(n 2
2 −n 3)
2 1/ 2

• TF • (1− LR ) • FC
m n1

assuming Lambertian source as before

m = demagnification factor
n1 , n2 , n3 = refraction indices of source medium, fiber core and cladding
TF = transmission factor
LR = loss due to Fresnel reflection

FC = fill factor of core

Example: m = 2; n1 = 1; n2 = 1.8; n3 = 1.5; TF = 0.8; FC = .85; and LR = 0

η = 15%
 Quantum efficiency of CCD

• QE = efficiency of system per photon incident on

receptor
• Absolute efficiency relative to light collection and
creation of electron within chip
• Not to be confused with DQE, though has influence
 Quantum efficiency influences

• Absorption of light
– Light needs to pass through polysilicon layers but not so
deep that cannot be captured in potential well
– Polysilicon layer needs to be relatively transparent
• Spectrum sensitivity
– Can modify this with overlaying materials and doping to
optimize to a given wavelength
• Light collection
– microlens
 Example Kodak Blue Plus CCD

• Indium Tin Oxide gate

• Microlens channels light to
ITO gate (more sensitive)
• Chip has a peak spectral
sensitivity in the range of 550
nm (flat over 575 to 650 nm)
• Matches the CsI peak
• quantum efficiency ~ 85%

Adapted from article by A. Ciccarelli, et al.,

Eastman Kodak Co.
ITO
 Noise

• Quantum noise inherent with exposure

• “shot noise” -- related to number of photons reaching
the CCD
• Dark current noise
– Strong dependence on temperature particularly at the
surface
– Thermal radiation can move electron to near the
conduction band, another can move it to the conduction
band
 Dark Noise Remedies

• Operate the chip in an inverted state

– Populate the interface states with free carriers
– Operate two or all three electrodes in inverted state
– Accomplished by doping
– Kodak uses this in the Blue Plus Chip mentioned earlier
• Over time dark noise will increase if readout has not
occurred
• Can use cooling (peltier, thermoelectric cooling)
– Especially needed for applications in astronomy
 Cooling Effects

• Dark noise may be of the order of 10,000 e/pxl/sec at

room temperature for unmodified CCD
• Drop of 40 degrees from room temperature will reduce
dark noise by a factor of ~ 100
• So is an important consideration
 Other noise sources

• noise associated with amplifiers in output stage

• Digitization noise
• Flat field correction
– Correct for non-uniform response across chip
– Dead pixels
– Application of correction adds to system noise
• Above common to any digital system but noise is
additive
 Implementations in Radiology

• Digital Fluoroscopy
• Stereo breast biopsy
• Mammography
• General radiography
 Digital Fluoroscopy

• Prevalent in II -- natural replacement for TV tube

– Output of II of the order of 1 inch which matches the size of a CCD
– Analogous readout to TV tube (data stream readout as lines)
• CCD has a linear response and a large dynamic range
– Blooming is greatly reduced and can be managed by manipulation of
the overflow pixel area of the chip
• Don’t have the problems associated with electron scanning
• Significant brightness gain afforded by II so losses in coupling less
of a problem
• Can operate at high frame rates
 Stereotactic Biopsy

• Typical FOV is 5 x 5 cm
• Matches with CCD well
• Fischer and Lorad use 2.5 x 2.5 cm cameras
– Fischer uses tapered fiber optics
– Lorad uses lens
• Noise is a major issue
– Use 27 to 28 kVp to increase photon flux reaching CCD
– 1024 x 1024 matrix; pixel size 50 µm -- affects noise
– CsI and special scintillators used to increase light output
 Stereotactic Biopsy

• Imaging characteristics
– DQE is low, unpublished values
– New cameras and scintillators have been implemented
– 512 matrix can help on contrast and noise
– but main reason for doing x-ray biopsy is to find
calcifications where resolution and noise is important
 Digital Mammography

• Original -- tiled arrays with fiberoptic tapers to cover 18 x 24

cm, not used now replaced by flat panel arrays
• Current -- Fischer
– Four rectangular CCDs
– Operating in time-delay-integration mode
– CsI scintillators to cover 1 cm x 22 cm area
– Scanned over 30 cm width
– Slot scanning affords significant scatter reduction
– No grid used so increase in flux helps overcome photon
limitations of CCD
 Time delay integration

• Good for long narrow area of moving data

• Parallel register is clocked in step with the motion
• When data reaches serial register, data are transferred
and stored in normal fashion
 General Radiography

• Desire 14 x 17 inch coverage

• Demagnification issues significant
• CsI scintillator common to increase efficiency
• Many different implementations
– Swissray
– Imaging Dynamics Corporation
 Swissray

• four CCDs with fiberoptic tapers

– Helps solve some of the demagnificagion issue
• 10% overlap of each CCD
– (stitching and balance of response)
 Imaging Dynamics

• single CCD
• Use a large high quality lens
• Have achieved high DQE, through many design features
 Specialized -- Statscan (Lodox)

• 12 CCDs coupled to GdOS

• Cover an array of 1 x 66 cm
• X-ray tube and receptor array attached to C-arm
• Assembly can move longitudinally up to 180 cm
• Angle up to 100 degrees
• CCDs operated in time-delay-integration mode
• Slot scan -- no grid needed; increases photon flux
• Pixel size 60 µm unbinned; different binning possible
• Small area resolution of 3.6 lp/mm, Whole body 1.7 lp/mm
 Advantages and Disadvantages

• Relatively simple • Demagnification is a major

• Cheaper to replace if failure concern

• Modularity -- easy upgrades • Relates to potentially lower

DQE
• Detector costs cheaper

• Vary with application

• II -- CCD a natural
replacement
• Stereotactic Biopsy --
demagnification less of a
problem
 Complimentary-Metal-Oxide Semiconductors
(CMOS)

• Are being used in place of CCDs

• Development less mature
• Size is an issue as with CCDs (demagnification)
• Readout scheme -- more direct output
– Each pixel is attached to a column and row output line
– Direct readout of a pixel
• Tiled arrays
• Specimen radiography, small area digital radiography
systems
 CMOS

• Uniformity correction may be tied to kVp for tiled

arrays
• Specimen biopsy radiograph system is set to a
particular kVp. If change kVp, see the tiling structure
unless one resets the “blank” calibration to that kVp.
• kVp range may be limited based on uniformity
 Conclusion

• Demagnification is a concern and much care needed

• Lower DQE compared to flat panel, in general
• Certain advantages
– Cost, modularity
• Based on cost and modularity, expected that systems
will continue to find use in Digital Radiography
• New designs are competitive