X-Ray Imaging Formation:

Analog and Digital Detectors

Introduction
Analog Detectors
Characteristics of Film
Optical Density
Contrast and Dose

Digital Radiography
Computed Radiography
Charged Coupled Devices
Flat Panel Arrays
Digital Image Processing
Digital Subtraction Angiography
Spring 2010

Radiation Detection & Measurements

Introduction
Analog devices as conventional X-ray radiographs
+ High resolution, simple, established, large areas, cheap
- Limited dynamic range, post-processing, portability, speed
Digital devices with advances in high-resolution pixel detectors
and digital signal processing
Computed radiography: Take analog (latent) picture first
and digitize later
Fully digital radiography: Take picture digitally
Digital detectors due to pixelation/ discretization of position
information; the signal produced in each pixel is an analog
packet of charges that is digitized by an analog-to-digital
converter (ADC) during image readout. Therefore, the initial
stage of digital detectors involve analog signals
Spring 2010

Radiation Detection & Measurements

Analog Detectors
Composition and Function
Film emulsion on top of flexible
sheet of plastic (Mylar): Grains
of silver halides (AgBr and
AgI) are bound in a gelatin
base
Cubic grain emulsion
Tabular grain emulsion

After exposure, latent

picture is imprinted (altered
chemical bonds)
Latent image is rendered
visible during film processing
by chemical reduction of the
silver halides into metallic
silver grains
Spring 2010

Radiation Detection & Measurements

Optical Density
X-ray film is negative recorder:
Increased exposure results in darker film
Degree of darkness is quantified by Optical Density (OD) measured
with a densitometer (localized light transmission)

Transmittance T (I0: intensity without film, I: intensity with film):

Spring 2010

Radiation Detection & Measurements

Hurter and Driffield Curve

Response of film to X-rays or light exposure
Unexposed film:
Film base OD ~ 0.11-0.15 (tinted)
Background (e.g. due to heat and
background radiation during storage):
Fog OD ~ 0.02-0.03
Base+Fog ~ 0.13-0.18
OD > 0.2 unacceptable
Film response to light from intensifying
screen is nonlinear/sigmoid
Hurter and Driffield studied this in
1890s (H&D or characteristic
curves)
Log-log plot of optical transmission
vs. X-ray exposure
Spring 2010

Radiation Detection & Measurements

H&D Curve II.

Contrast:

Gamma: maximum slope

at inflection point of
characteristic curve

Slope contrast
Average gradient:
(OD2-OD1)/(log(E2)-log(E1))
Film contrast controlled by size of
distribution of silver grains
Larger slope w/ high contrast results in
reduced latitude, i.e. dynamic exposure range

Speed:
Sensitivity or Absolute Speed of screen-film
can be deduced from H&D curves:
[Exposure(OD=1)]-1
Speed is commonly used as relative measure
in so-called Par-speed system defined with
Speed=100, assuming a Speed=200 system to
be twice as fast.
Rare-earth based screen-film combinations
result typically in Speed=400
Higher speed allows to achieve a specific OD
level with less exposure, i.e. less radiation=
number of photons, however, potentially
introducing more statistical noise!
Spring 2010

Radiation Detection & Measurements

Digital Radiography - Concept

Computed Radiography (CR) Analog detectors, digital storage
Based on photostimulable phosphor
detector (PSP) systems
X-ray energy is trapped in PSP screen to
be read out later (storage phosphors or
imaging plates)
Introduced in 1970s, increasing use in
1980s in parallel to increased electronics
record keeping (PACS: Picture Archiving
and Communication System)
CR is composed of flexible screen made
of BaFBr (85%) + BaFI (15%) (+ small
quantity of Eu) enclosed in a cassette
(similar to screen-film cassette)
Latent or stored image after exposure
is read out by a PMT after light emission
stimulated by a localized laser beam
Spring 2010

Radiation Detection & Measurements

Computed Radiography - Exposure & Readout

Stimulable phosphor:
Excitation of electrons from VB to CB and
trapping of some electrons in metastable F
-centers
Release of electrons in F-center through
stimulation (small energy = red laser) and
transition to VB by emitting light (larger
energy difference = blue light)
Light intensity emitted by the imaging plate
is proportional to absorbed X-ray intensity

Readout:
Localized and moving laser beam stimulates
position sensitive blue light emission
captured in PMT (position is translated into
time)
Not all trapped electrons are released in one
scan; several scans can be performed w/o
significant degradation.
Very bright light source resets most of the
trapped electrons enabling the plate to be
used2010
again.
Spring
Radiation Detection & Measurements

Computed Radiography Dynamic Range

CR system can be operated in a
very similar way as screen-film
radiography
Advantage of PSP/CR is much
larger dynamic range (latitude)
Almost no under- or over
-exposure resulting in a
necessary retake

Spring 2010

Radiation Detection & Measurements

Digital Detectors
Flat-panel detectors (~30-40cm2) under development for over 30 years
(x2 and x3 detection efficiency over CR and screen-film, respectively)
Direct vs. indirect conversion detectors
Direct:
Conversion of X-ray energy
into electrical signal
(reflecting the exposure) in
photoconductor (amorphous Se
or Si, CdZnTe, or PbI2) and
position-sensitive readout

Indirect

Direct

Indirect:
Scintillation material (such as
amorphous Si or CsI) converts
energy of X-ray photons to
visible light
Photoconductor converts
visible light into electrical
signal (CCDs or amorphous Si
coupled to TFT arrays)
Spring 2010

Radiation Detection & Measurements

10

Digital Detectors: Charge-Coupled Devices

First developed at Bell labs in 1969,
now widely available in video- and
photo-cameras
CCD itself is integrated circuit made of
crystalline Si with discrete pixel
electronics etched into its surface (up
to ~16Mpixels, e.g. 4x4cm2 with 10m
pixel sizes)
Si surface is photosensitive, i.e. visible
light liberates electrons
Electrons are kept in each pixel by
electronics barrier produced by
external voltage during exposure
After exposure the CCD is read out
similar to a bucket brigade
CCD cameras produce high-quality
images from visible light
Normally used in combination with
focusing lens or light coupling optics
Spring 2010

Radiation Detection & Measurements

11

Flat Panel Detectors - Concept

Pixel detector with horizontal and vertical control
and readout lines (e.g. for 1000x1000 array 2000
lines instead of 106 readout channels)
Similar to computer monitor, here read signal, not
send signal for display
As in CCD, each pixel stores charge during exposure
A transistor (+ associated electronics arranged in
the corner of each pixel) stores the charge (source)
Control: -V on gate lines R during exposure
Readout: Raster readout
+V sequentially to each gate line R;
Multiplexer sequentially connects each vertical
signal line C via switches S to the digitizer allowing
each pixel along one row to be read out
Therefore, gate line selecting the row and the
multiplexer selecting the column to be read
In contrast to CCD, signals/ charges are not passed
through other detector elements but separate
signal lines (transfer efficiency ~98% vs. 99.99% in
CCD)
Spring 2010

Since device is manufactured in thin-film

deposition technology, these flat-panel system
are called thin-film transistor (TFT) image
receptors

Radiation Detection & Measurements

Gate
Source
Drain

12

Flat Panel Detectors Fill Factor

Size of pixels largely determines
spatial resolution of detector
system (typically 100 m)
To improve resolution, smaller
pixel sizes are required
Since readout electronics takes
fixed space, a larger fraction of
the area is not sensitive to light
and the light collection
efficiency decreases
Trade-off: Spatial resolution vs.
contrast resolution

Spring 2010

Radiation Detection & Measurements

13

Flat Panel Detectors

Indirect vs. Direct Detection
Indirect:
Scintillation X-ray-to-light conversion
Direct:
Photoconductor material on top of
TFT
Advantage of direct detectors:
Improved spatial resolution
No blurring due to isotropic light emission /
electrical field lines confine electrons to
straight path towards readout

Improved conversion efficiency

Electrical field lines can be shaped to guide
electrons to sensitive areas only

Lack in low-Z (Se, Z=34 vs. Gd/Cs,

Z=64/55) can be compensated with
larger thickness or new materials
(CdTe, HgI2, PbI2)

Spring 2010

Radiation Detection & Measurements

14

Digital Image Processing - Artifacts

One of the main advantages of a
digital image is possibility to modify
its appearance after the picture
was taken.
Artifacts due to dead pixels can be
corrected (by averaging)
Background / electronic noise can
be measured for each pixel and
subtracted [using dark image
D(x,y)]
Differences in pixel efficiencies and
electronics gains can be corrected
[using gain image G(x,y)]

Spring 2010

Radiation Detection & Measurements

15

Global Processing

Very wide window

(low contrast)

Application of simple function to all pixels:

I(x,y) = b[I(x,y)-a]
Limits can be placed on I(x,y): Thigh, Tlow
If I>Thighthen I=Thigh
If I<Tlowthen I=Tlow
Examples:
Reversing contrast (-b)
Windowing (Thresholds)
Leveling (a)

Spring 2010

Radiation Detection & Measurements

Narrow window
(acceptable contrast)

Contrast inversion

16

Convolution I.
Important mathematical operation to
manipulate digital images:

Example with 3x3 kernel

I(x): Input
g(x): result
h(x): Convolution kernel

Essentially, shifting and adding gray

-scale values in discrete domain
Examples:
Delta function Shifting - Smoothing
0

Spring 2010

Radiation Detection & Measurements

17

Convolution II.
Examples:

Original Image

Blurring /smoothing with

11x11 convolution kernel)

Edge-enhanced image

Harmonization

Blurring/ smoothing/ low-pass

filter
1/9 1/9 1/9
1/9

1/9

1/9

1/9

1/9

1/9

Edge enhancements /sharpening

-1

-1

-1

-1

-1

-1

-1

-1

Harmonization
Original - blurred image resulting in
reducing low-frequency and
emphasizing high-frequency
structures

Deconvolution reverses convolution

except for noise penalty
Spring 2010

Radiation Detection & Measurements

18

Effects on Image
Impact of
Finite pixel size

# of pixels per
image

1024x1024

64x64

32x32

16x16

Dynamic range

# of bits per pixel

Spring 2010

Radiation Detection & Measurements

1
19

Digital Subtraction Imaging I.

Digital Subtraction Angiography (DSA)
Most common example of temporal
subtraction methods:

Digital radiograph is acquired representing

the anatomy (= mask)
Injection of radio-opaque contrast agent
Digital radiographs are acquired during and
after injection
Mask image is subtracted from images
containing contrast agents

DSA results in excellent detail of

vascular anatomy while other
components are subtracted out
DSA allows quantitative
measurements, e.g. of stenosis
(based on vessel area)
Spring 2010

Radiation Detection & Measurements

20

Digital Subtraction Imaging II.

Dual Energy Subtraction
Exploit differences of
bone (Zeff~13) and soft
tissue (Zeff~7.6) due to
different Z
Based on digital projection
radiography taken at two
energies

I=ln(I1)-R ln(I2)

Spring 2010

Radiation Detection & Measurements

21

Digital Subtraction Imaging III.

Dual Energy Subtraction

Low-energy (56 kV)

High-energy (120 kV)

DES: Boneweighting

DES: Tissue
weighting

Exploit differences of
bone (Zeff~13) and soft
tissue (Zeff~7.6) due to
different Z
Based on digital projection
radiography taken at two
energies

I=ln(I1)-R ln(I2)

Spring 2010

Radiation Detection & Measurements

22