Mid Term Study Guide

Scintillator layer of indirect FP: a burst of light and effects on spatial resolution

The burst of light within the CsI layer is more likely to occur in the anterior portion of the phosphor, this means the light produced must travel a relatively large distance before exiting the Phosphor layer and striking the TFT array. Because the light bursts in all directions, and travels some distance, there can be appreciable blurring as the light is read by multiple adjacent dexels.  

Why do we need a stationary grid in the Fluoroscopy C-arm unit?

The grid is necessary for improving contrast by reducing detected scatter radiation. Because Fluoroscopy images are acquired in real time, a moving grid would create motion artifacts.

Can be removed for low dose procedures

Detector Pixel structure (detector element)

Indirect detection: CsI phosphor layer converts x-rays to light, Thin Film Transistor (TFT) array, Capacitor, and charge collection electrode…. The light created in the scintillator is collected by the electrode and stored in the capacitor. The TFT is a switch with 3 connections: gate, source, drain. The TFT is closed during exposure, allowing charge collection at each dexel. The TFT is sequentially opened - row by row - after exposure, allowing the charge in each dexel’s capacitor to drain outside of the panel area to an amplifier.

Direct detection: A semiconductor, typically amorphous selenium (a-Se) produces electron-hole pairs proportional to incident x-rays, which are directly converted to charge. The semiconductor is layered between 2 surface area electrodes connected to bias voltage and dielectric layer. The dielectric layer prevents overcharging the dexels which could damage the TFT array.

What is the typical size of a detector element? Why is this important?

~200μm (CHECK THIS) The entire dexel area is not photosensitive, due to the integrated electrical components. Larger dexel sizes allows for a higher percentage of photosensitive area on a dexel, the fill factor. The 200μm dexel has an 80% fill factor, while the smallest dexel, 100μm, has a fill factor of ~40-50%.

Larger dexels may also be necessary, despite the drawback of poorer spatial resolution, because the bandwidth may restrict the amount of image data that can be displayed at 30 FPS. Binning 2x2 dexels effectively reduces that amount of data transfer by a factor of four.

What happens if we don’t follow the Nyquist sampling theorem?

Aliasing.. (Anything else?)

What is Aliasing?

a.      A.k.a. “wrap-around”. Occurs when a high frequency signal is sampled at a rate lower than the Nyquist frequency. High frequency signal is recorded as low frequency (aliased) the high frequencies are plotted at the low end (wrap-around)..

b.      Appears as a Moire pattern or wavy lines. Can be caused by anti-scatter grid

c.      Can be fixed in MRI by increasing FOV. Can be fixed in US by changing velocity range (commonly seen in Doppler mode)

Structure of third CT generation; why this model is successful?

It is stable, easy to maintain, and produces least amount of artifacts. The fourth gen’s full bore detector array meant difficulty repairing and couldn’t use collimator at the detector. (ANYTHING ELSE)

Two factors affecting image resolution and image contrast in CT images.

a.      Spatial Resolution: Detector Width, Convolution Kernel Filter, Slice Thickness

b.      LCD: Kernel Filter, mAs, Slice Thickness

Central slice theorem and how to apply to CT reconstruction.

a.      CST is the 1D FT of a projection at an angle is equal to the profile through (slice) the 2D FT at the same angle. CST is the reconstruction algorithm; the CT equivalent of k-space in MRI.

How to acquire slice thickness in CT imaging?

Multiplies of detector width

Where to apply kernel filtering in CT?

a.      Fourier Space because convolution here is a simple multiplication, and faster to do here 9Bushberg, 354). Fourier space is organized in spatial frequency, kernel multiplication here will directly filter the data to be 2D INVERSE Fourier Transformed into the image. (CHECK THIS)

Examples of Kernel filtering in CT

a.      Higher FC is for sharper images, better edge definition (bones) and spatial frequency, but higher noise. Higher noise because the FC amplifies higher frequencies, and noise is high frequency. Decreases signal because noise overcomes.

b.      Lower FC improves contrast because it amplifies the signal and reduces noise

c.      PICTURES OF FC GRAPHS.

What is sinogram?

a.      Raw projection data visualized over all angles. (ANYTHING ELSE)

How to quantify noise in CT imaging?

a.      Noise is calculated as the standard deviation of pixel values within an ROI. This standard deviation can be represented in HU by propagating the noise across an ROI HU: σ=1000(σ_x,y,z - σ_water)/(σ_water)

b.      This calculation of noise in HU is only possible because the signal and noise both come from the same source!

How is noise affected by a change in kV and mAs (study the example in your handout)

Inversely proportional to the sqrt(dose)

and

Inversely proportional to the sqrt(ST x mAs)

LOWER kV increases quantum noise because there is less fluence

How is S/N affected by changes in slice thickness, kV and mAs (study the example in your handout)

a.      Thinner slices increases contrast and Spatial Resolution because less scatter from a wider field, but increases noise because less active detector area – less photons detected means more noise. This can be mitigated by increasing dose, mAs, or the amount of signal reaching the detector. Slice Thickness and mAs together determine SNR…

SNR proportional to sqrt(mAs x ST)

increase kV increases SNR because higher fluence

What is the Hounsfield Unit (HU) scale?

a.      Unit of measure representing electron density of different materials. (ANYTHING ELSE?)

Factors affecting Spatial sampling and temporal Sampling in CT

a.    Minimum sampling rate is determined by Nyquist Frequency, , where D is the detector width – SPATIAL SAMPLING.

b. Temporal sampling is done by the ADC which converts Voltage to binary

Factors affecting CNR in CT images

a.      Slice Thickness, thinner slice is better CNR because there is less scatter from a wider field, BUT MORE NOISE because smaller detector area receiving signal.. higher electronic noise. (MORE)

b.      mAs. Higher mAs means signal reaches the detector, which reduces the amount of noise in the sampled data. Higher CNR.

c.      Kernel Filter (FC), lower FC preferentially boosts low frequency signal while reducing high frequency intensities. Reducing high frequency data reduces noise, improving contrast but decreasing spatial resolution…

d.      Higher sampling rates improve LCD by providing more signal data to be digitized.  

Basic of Fourier Transform (FT) and Convolution Integration

a.      FT is an algorithm that breaks down a signal frequency into the all of its component frequencies and their respective intensities.

b.      Convolution integration is the moving average of two functions in time.

Why we need sinogram in CT imaging?

It simplifies the reconstruction process by organizing the projection data from all angles into a single data set.

What is Helical Pitch in CT?

The relative speed of the table compared to the rotation of the gantry. Higher pitch creates a more spaced out helix.

HP = table speed (mm/rot) / detector config.

HP < 1 increases dose because rotations overlap

Why do we use a Helical CT scan?

a.      Helical exams are faster because they eliminate the start/stop time required for axial scans to capture slices.

b.      Less dose because faster and less volume exposed to field (assuming HP > 1)

Image quality in MR image acquisition: meaning of T1 and T2 weighted images

a. M_T = M_o(1 - e^-TR/T1)e^-TE/T2

b.      T1 Weighted: Small TE & TR (10-30 ms, 400-600 ms). Emphasizes the difference in longitudinal relaxation times.

                  Adipose tissue is intensified, fluid structures (e.g. CSF) are hypointense.

c.      T2 Weighted: Long TE & TR (90-110 ms, 1500-3000 ms). Emphasizes the transverse relaxation times.

                Fluid structures  (CSF) are intensified

Source of Noise in MRI

a.      MOST noise is produced because of the conductivity of human beings. RF pulses ≥ 1MHz (assuming a cooled coil) will generate noise from the tissue.

b.      At the system level, noise may be produced by active electronic components, such as PIN diodes, or inadequate isolation between the Transmitter and Receiver channels.

Conditions to select slice in MRI: RF pulse/ Gradients field and slice thickness

a.      Must use shaped RF pulse to specify the frequency bandwidth, and to deliver uniform power along that bandwidth.

b.      Gradient field sent at same time as RF pulse

Principle of Phase encoding and frequency encoding in MR images

a.      Phase encoding (Gy) gradient provides spatial frequency in the Y-direction of the selected slice (from Gz). The more Gy pulses for a selected slice with give better spatial resolution by building more data in k-space.

The frequency selection (Gx) gradient provides data in the x-direction corresponding to the row of k-space built by Gy. (CHECK THIS)

Different between T2 and T2* relaxation time

T2* is faster than T2 because it accounts for reduced relaxation time caused by field inhomogeneity

1/T2* = 1/T2 + γΔB

Factors affecting spatial resolution in MRI

a.      Number of Gy repetitions determines the image matrix size

b.      FOV (ADD MORE HERE)

Factors affecting low contrast in MRI

T1 & T2 weighting will determine how long TE and TR are

How do we specify the main magnetic field non-uniformity in MRI?

4 tests to describe ppm of the B-field strength over a spherical volume (DSV)…

1. Spectral Peak converted to ppm from Larmor Equation

   1. FWHM (ppm) = FWHM (Hz) / 42.58B_o

2. Bandwidth Difference test only accesses field inhomogeneity in the direction of frequency encoding axis à need more images with frequency encoding across all axes

3. Phase Map test can provide a pixel-by-pixel measurement of inhomogeneity, provided the MRI system can display phase-contrast images.

4. Phase-Difference Map is identical to the above, but with an added image set for longer TE… This test yields greater sensitivity at the expense of time

How to compute Percent Image Uniformity (PIU) in MRI?

PIU = 100 x (1 - (ROI_Max - ROI_Min) / (ROI_Max + ROI_Min))

1.5T: 87.5%....3T: 82%... Slice in center of phantom free of internal structures.

Compute CNR?

CNR = (HU_in - HU_out) / σ_out

ACR minimums: Head > 1.0, abdomen > 0.7

Water and air HU: expected values and acceptable limitations

For ACR QC phantom…

a. Water should be 0 ± 5 HU, but MUST be 0 ± 7 HU

b.      Air is -970 to -1005 HU

Factors affecting water and air HU values?

a.      kVp. Higher kVp reduces beam hardening effect. Lower kVp will be most affect HU values due to the larger portion of energy spectrum that is attenuated in the filter.

b.      Scatter can contribute unwanted signal, artificially increases HU values

c.      FC choice, specifically higher FC values, can shift HU values

How to preserve linearity of CT system when applying convolution filter?

Kernel filter in frequency domain must be symmetrical

Why is MRI practical application of Fourier Transform?

The MRI signal (echo) is acquired in time, however, it contains spatial frequency information of all objects within the selected slice

How is rotating reference frame acquired?

Subtract resonant frequency

Why is MRI inefficient?

Difference in anti-parallel and parallel spin populations is only ~5 ppm… produces signal on order of MICROvolts

What is coupling factor?

Efficiency of energy transferred from RF coil to patient. How effective the B field is at flipping nuclei. High coupling factor means stronger signal

Why use 90^o flip-angle in pulse sequence?

Provides largest possible rotation of the aligned nuclei into the transverse plane (M_xy) which produces the largest detectable signal

ACR minimum for LCD?

a.      1.5T : 9 of 40 spokes resolvable

b.      3T : 37 of 40 spokes resolvable

What does MTF tell us about the system?

MTF determines the system’s resolution by evaluating how contrast diminishes with increasing signal frequency.

What is Nyquist sampling Theorem and why do we need it?

Minimum sampling rate required to obtain accurate image reconstruction.

Main cause for PIU failure?

RF non-uniformity. Must be calibrated but can be helped using multi-channel head coil

Why does MRI only need one FT (inverse) but CT needs two?

MRI is actual application of FT… MRI produces spatial frequency, CT has to transform from projected to spatial and then to image.

Why is there no sinogram in MRI?

MRI is actual application of FT… MRI produces spatial frequency, there is no projected data

What is SAR?

Specific Absorption Rate, heat felt by patient from the RF coil. Regulated by FDA.