NMT Boards – Image Processing & Informatics

Predominant photon interaction in soft tissue at nuclear medicine energies

Compton scatter

Interaction in which a photon gives ALL its energy to an inner-shell electron

Photoelectric absorption

Interaction responsible for full-energy photopeak counts in the crystal

Photoelectric absorption

Minimum energy needed for pair production or positron decay

1.02 MeV

Z and A change in positron decay and electron capture

Z −1, A unchanged

Decay mode used by proton-rich nuclides with less than 1.02 MeV available

Electron capture

Energy and angle of the two annihilation photons

511 keV each, ~180° apart

Two decay modes that produce the SAME daughter

Electron capture and positron decay

Percent error (1 SD) formula for N counts

100 ÷ √N

Counts needed for 1% error

10,000 counts

Percent error for 40,000 counts

0.5%

Confidence level for ±2 SD

95%

95% range for an average of 12,555 cpm

12,331–12,779 cpm

Counting-time formula for a desired percent error

[(Rs + 2Rb) ÷ Rs²] × (100 ÷ v)²

Test used to detect systematic counting error

Chi-square test

Type of error that can be predicted but never eliminated

Random error

Precision means

Repeatability (low random error)

Accuracy means

Correctness (no systematic error)

On semilog paper, exponential decay plots as

A straight line

Instrument best for finding contamination

GM survey meter

NRC-required survey meter reading range

0.1–100 mR/hr

Instrument for ACCURATE exposure rates from therapy patients

Portable ionization chamber (cutie pie)

Instrument used for wipe tests and blood/urine samples

Well counter

Activity above which the well counter shows dead-time losses

~2 µCi (74 kBq)

Collimator that gives better resolution but lower sensitivity

LEHR (high-resolution)

Collimator that is a compromise of resolution and sensitivity

LEAP / GAP (all-purpose)

Collimator type that magnifies the image

Converging

Collimator type that minifies the image

Diverging

Effect of increasing source-to-collimator distance on parallel-hole resolution

Resolution decreases

Tc-99m 20% energy window range

126–154 keV

In-111 245 keV 20% energy window range

~221–269 keV

Tl-201 energy windows (Bolus Ch. 20)

30% at 70 keV + 15% at 167 keV

Maximum counts per pixel in byte mode

256 (8 bits)

Maximum counts per pixel in word mode

65,536 (16 bits)

What happens when a pixel overflows

Extra counts are lost (inaccurate data)

Memory needed for word mode vs. byte mode

Twice as much

Memory for a 128 × 128 word SPECT with 64 projections

~2 MB (2,097,152 bytes)

Pixel size formula

FOV ÷ (matrix × zoom)

Pixel size for 500 mm FOV, 512 matrix, zoom 1.0

0.97 mm

Rule of thumb for pixel size relative to system FWHM

≤ one-third of FWHM

Does post-acquisition zoom improve resolution

No (only enlarges)

Requirement for the uniformity flood when using SPECT zoom

Acquire flood at the same zoom

Acquisition mode synchronized to the ECG R-wave

Gated

Acquisition mode storing each event as x, y and time

List mode

Main advantage of list mode

Retrospective reframing/regating

Bolus renal dynamic frame rate example

20 s/frame × 25 min (75 frames)

HIDA dynamic framing

1 frame/min for 60 min

Artifact caused by simple back-projection

Star (spoke/streak) artifact

Filter that removes the star artifact in FBP

Ramp (high-pass) filter

Major limitation of filtered back-projection

No attenuation model; amplifies noise

Iterative algorithm using ALL projection data each iteration

MLEM

Iterative algorithm that splits data into subsets

OSEM

OSEM equivalent MLEM iterations

Iterations × subsets

Requirement for the distance from the center of rotation

Must remain constant

Format PET data are stored in (LOR by angle and position)

Sinogram

Sensitivity gain of 3-D PET over 2-D

~5×

Scatter fraction in 3-D PET

~30–40%

Butterworth filter's dominant control

Cutoff frequency

Effect of lowering a filter's cutoff frequency

Smoother/blurrier image

Number of parameters in a Hanning filter

One (cutoff)

Filter type of Butterworth and Hanning

Low-pass

Nyquist frequency in cycles/pixel

0.5 cycles/pixel

Appearance of an over-filtered image

Mushy/blurry; lesions hidden

CT windowing maps what to gray levels

Hounsfield units

PET coincidence timing window

~4–12 ns

PET randoms delayed window

60–100 ns

Term for PET localizing events without a lead collimator

Electronic collimation

PET crystal with highest attenuation but slow and low light

BGO

Crystals fast enough for time-of-flight

LSO and LYSO

PET spatial resolution vs. SPECT

~4 mm vs. ~12 mm

PET sensitivity relative to a gamma camera

50–100× more sensitive

Axial FOV per PET bed position

15–18 cm

What time-of-flight PET does

Localizes the event along the LOR

Prompt coincidences consist of

Trues + randoms + scatters

NECR formula

T² ÷ (T + S + kR)

NECR equals the square of what

Signal-to-noise ratio

Optimal injected activity relative to NECR

At or below peak NECR

Random coincidences rise with

Activity² and 3-D mode

Coincidence where a photon was Compton-scattered onto a wrong LOR

Scatter

Are PET randoms corrected by the energy window

No (delayed window or singles)

Daily PET QC

Blank scan and PMT baseline/gain check

Quarterly PET QC

Normalization and cross-calibration

Minimum duration of a PET normalization scan

≥6 hours

PET calibration converting pixel values to activity/mL for SUV

Absolute activity (cross) calibration

Thin diagonal line on a PET blank-scan sinogram

Single faulty crystal

Broad diagonal band on a PET blank-scan sinogram

Block/PMT (module) failure

Which is acquired first in PET/CT

CT

X-ray production process in the CT tube from braking electrons

Bremsstrahlung

Low-dose CT kVp range

80–140 kVp

Low-dose CT mAs range

20–120 mAs

Effect of raising kVp on CT

Less noise, more dose, less contrast

CT pitch formula

Table travel per rotation ÷ (slices × slice width)

Pitch for 40 mm table travel with 16 × 2.5 mm slices

1.0

Effect of a higher reconstruction kernel number

Sharper image

CT noise from too few x-rays

Quantum mottle

Hounsfield unit formula

1,000 × (µtissue − µwater) ÷ µwater

HU of water

0

HU of air

−1,000

HU for µtissue 0.21 and µwater 0.19

~105 HU

CT window width controls

Contrast