HST.583 Functional Magnetic Resonance Imaging: Data Acquisition and Analysis Fall 2006

Transcription

1 MIT OpenCourseWare HST.583 Functional Magnetic Resonance Imaging: Data Acquisition and Analysis Fall 2006 For information about citing these materials or our Terms of Use, visit:

2 HST.583: Functional Magnetic Resonance Imaging: Data Acquisition and Analysis, Fall 2006 Harvard-MIT Division of Health Sciences and Technology Course Director: Dr. Randy Gollub. INTRODUCTION This Lab consists of three parts: Part : SNR Measurements Temporal and Spatial Characteristics in Signal and Noise Part 2 : Determination of MR parameters (T, T2, T2*) across tissue types and regions of interest. Part 3 : EPI distortions due to B0 Inhomogeneity All experiments will be performed on a human subject. The SNR measurements will also be run on a phantom for comparison. Some of the data analysis will be performed on the scanner console, however you will be asked to note the measurements obtained as you will need them to solve the exercises given in the lab report. The main goals of this lab are to: ) Become familiar with basic principles of MRI Physics and measurements (i.e. SNR, relaxation times, etc). 2) Understand the T, T2 and T2* properties of various tissue compartments. 3) Acquire and evaluate phantom data. 4) Perform a human scanning experiment and investigate the various sources of noise in the fmri time series. 5) Evaluate EPI distortions through field maps and by varying the readout properties.

3 . SNR Measurements Background At the most basic level, the SNR depends on the number of protons present in the voxel (proportional to Voxel Volume if we assume a constant density) and the MeasurementTime. The latter is a standard aspect of signal averaging assuming the noise is uncorrelated and distributed in a Gaussian distribution. Each acquisition in the k- space matrix is essentially averaged when the Fourier Transform produces the image. Therefore the total measurement time is the total amount of time the digitizers are actually recording k-space samples. For a readout line of 256 samples acquired with a dwell time (time per sample) of 25μs, this yields an acquisition time of 6.4ms. Sometimes acquisition time for the readout is expressed in terms of the bandwidth of frequencies present across the image ( BW read ). BW read = equal to 40kHz for this example. dwell time the Siemens system the BW is expressed in Hz per pixel across the image, so for the 256 matrix above, this is a BW read = 56 Hz/px. The total image acquisition time is the time per line multiplied by the number of lines (# phase encode steps N PE ) and the number of times each line was averaged (N AVG ) Equation shows the dependence of SNR on some of the above parameters: Voxel Volume N SNR AVG () BW read For fmri, it is the time-course SNR that is important. Functional MRI is restricted by multiple sources of variance, such as instrumental sources of error including thermal noise and shot-to-shot electronic instability, and subject dependent modulations of the MR signal associated with physiological processes. In addition to respiratory and cardiac cycle contributions, the physiological noise also consists of a noise element with BOLD-like TE dependence (Triantafyllou et al. 2005), (Krueger and Glover 200), and spatial correlation within gray matter (Krueger and Glover 200). The origin of this BOLD noise is still not fully understood, but is generally associated with hemodynamic and metabolic fluctuations in the gray matter. Since the physiological fluctuations represent a multiplicative modulation of the image signal (Krueger and Glover 200) their amplitude scales with the MR image intensity. This is in contrast to the thermal noise sources which can be represented by the addition of a fixed amount of Gaussian noise power whose amplitude is determined primarily by the coil loading. If the noise sources are assumed to be uncorrelated, the total noise in the image time-course (σ) is related to its thermal (σ 0 ) and physiological (σ p ) components via: σ = σ + σ, (2) p In our measurements, shot-to-shot scanner instabilities will contribute to both terms, σ 0 and σ p, depending on their signal dependence. Phantom measurements, however, show that they comprise only a small fraction of the in vivo time course noise. The time-course SNR (tsnr) is then defined as:

4 tsnr = S, (3) 2 σ 0 + σ 2 p where S is the mean image signal intensity. Defining the SNR in an individual image as S SNR0 = and combining with Eq. (3), we determine the relationship between tsnr σ 0 and SNR0: tsnr = SNR0 (4) + λ 2 SNR0 2 where λ is a constant. References. Triantafyllou C., et al., Comparison of physiological noise at.5 T, 3 T and 7 T and optimization of fmri acquisition parameters. NeuroImage 26, ; Krueger G, Glover GH. Physiological noise in oxygenation-sensitive magnetic resonance imaging. Magn Reson Med 46: 63-7; 200.

5 Experiments In this exercise we will acquire human MRI data in order to characterize image signal and noise characteristics and the time-course signal and noise characteristics. Data will be collected on a 3T Siemens Imager. We will examine the spatial SNR, time-course SNR and their relationship for three different image resolutions, 5x5x5mm 3, 3x3x3mm 3 and.5x.5x.5mm 3. In all cases, images with zero RF will also be obtained to capture the thermal image noise. For comparison, time series data using the same parameters will also be acquired on a loading phantom. Acquisition: ) Localizer. 2) EPI time series at low resolution 5mm x 5mm x 5mm, 9 slices, 50 time points, TR=2000ms, TE=32ms, (see protocol epi_5x5x5_signal). 3) Same as #2 but with no RF excitation (just thermal noise) 4) EPI time series at medium resolution 3mm x 3mm x 3mm, 9 slices, 50 time points, TR=2000ms, TE=32ms (see protocol epi_3x3x3_signal). 5) Same as #4 but with no RF excitation (just thermal noise). 6) EPI time series at higher resolution.5mm x.5mm x.5mm, 9 slices, 50 time points, TR=2000ms, TE=32ms (see protocol epi_.5x.5x.5_signal). 7) Same as #6 but with no RF excitation (just thermal noise). Note : Typically, thermal noise would be calculated by drawing an ROI outside the signal area in an image. However in EPI acquisition there are a lot of artifacts present. To avoid misreading the noise we thus acquire a separate image without an RF that provides a better representation of thermal noise. Note 2: Since the thermal noise is random we need to characterize it in terms of its mean, and standard deviation (or variance). Before we can calculate these quantities, we also need to know what kind of statistical distribution this noise belongs to. For example, the most common type of statistical distribution is the Gaussian or normal distribution but the spatial MRI noise outside of the brain has been empirically determined to follow a Rayleigh distribution. It is thus simple to compute the mean and variance of the thermal noise by first computing its variance and mean as though it were Gaussian and applying a Rayleigh correction factor to account for this difference. Spatial SNR (SNR0) The SNR in an individual image (SNR0) is a measure or the image quality. In our experiments we will evaluate the impact of the spatial resolution on the SNR0. In human data, ROIs will be defined in cortical gray matter. The SNR0 for a given pixel will be calculated as the mean pixel value for all the images in the time-series divided by the standard deviation of the thermal noise of the time-series acquired with no RF excitation (zero flip angle images).

6 . Load the EPI images and the Noise time-courses on the mean curve task card. 2. Select the EPI time-course, draw an ROI within the signal area, and record the mean signal value. 3. Select the Noise time-course, draw an ROI and record the standard deviation. 4. Record your measures in Table and calculate the SNR0. 5. Repeat steps -4 for all three spatial resolutions. 6. Repeat steps -4 for the phantom data at all three resolutions and record results on Table 2. Lab Question : Draw the calculated SNR0 as a function of voxel size and comment on your findings. Describe the differences if any, between the human and phantom data. Temporal SNR (tsnr) Temporal SNR is defined as the image-to-image variance in the time-course and will be measured on a ROIs based analysis in the cortical gray matter. Temporal SNR in a given pixel will be determined from the mean pixel value across the 50 time points divided by its temporal standard deviation.. Load the EPI time series images on the viewer. 2. Calculate the Standard Deviation map from the EPI time series through the scanner UI. Open the Patient Browser, go to Evaluation -> Dynamic Analysis -> Standard Deviation. Press within series, test and assign a name to the new image (STD_mymap). A new series is created on your patient browser, named STD_mymap. 3. Load the images on the mean curve task card. Select both the EPI time course and the standard deviation map and draw an ROI within the signal area. 4. Record the mean value within the ROI on the EPI images; that is your temporal mean of the signal. 5. Record the mean value within the ROI on the standard deviation map; that is your temporal noise. 6. Calculate the temporal SNR from the above quantities and note on Table. 7. Repeat steps -6 for all three spatial resolutions. 8. Repeat steps -6 for the phantom data at all three resolutions and record results on Table 2. Lab Question 2: Draw the calculated tsnr as a function of voxel size and comment on your findings. Describe the differences if any, between the human and phantom data. Relationship between SNR0 and tsnr The tsnr will be analyzed as a function of SNR0 for the given set of resolutions. Use the recorded values from Tables and 2 incorporating the model for tsnr from Eq. 4 (where λ=0.007).

7 Lab Question 3 : Show the relationship of tsnr as a function of SNR0 when SNR0 is modulated by the voxel size. Describe the differences, if any, between the human and phantom data. What is the asymptotic limit for tsnr? Lab Question 4 : You are asked to perform an fmri study of medial temporal lobe activation at a high field strength. Which acquisition parameters would you consider most important to optimize in order to achieve the best activation results? For a 3T scanner provide a suggested set of acquisition parameters. Lab Question 5 : Draw ROIS on various tissue types, generate the tsnr as a function of SNR0 in gray matter, white matter and CSF. Record mean signal and standard deviation of the noise. Comment on your findings. Table Human Data Average values for SNR measurements as a function of image resolution. SNR0 corrected for Rayleigh distribution. Signal Thermal Time-Series Spatial Temporal (mm 3 ) Noise Noise SNR SNR.5x.5x.5 3x3x3 5x5x5 Table 2 Phantom Data Average values for SNR measurements as a function of image resolution. SNR0 corrected for Rayleigh distribution. Signal Thermal Time-Series Spatial Temporal (mm 3 ) Noise Noise SNR SNR.5x.5x.5 3x3x3 5x5x5

8 Table 3 Human Data - Gray Matter Average values for SNR measurements as a function of image resolution. SNR0 corrected for Rayleigh distribution. Signal Thermal Time-Series Spatial Temporal (mm 3 ) Noise Noise SNR SNR.5x.5x.5 3x3x3 5x5x5 Table 4 Human Data - White Matter Average values for SNR measurements as a function of image resolution. SNR0 corrected for Rayleigh distribution. Signal Thermal Time-Series Spatial Temporal (mm 3 ) Noise Noise SNR SNR.5x.5x.5 3x3x3 5x5x5 Table 5 Human Data - CSF Average values for SNR measurements as a function of image resolution. SNR0 corrected for Rayleigh distribution. Signal Thermal Time-Series Spatial Temporal (mm 3 ) Noise Noise SNR SNR.5x.5x.5 3x3x3 5x5x5

9 2. T, T2 and T2* Measurements Background To calculate T values of the tissue we will be using an inversion recovery turbo spin echo sequence. We will be sampling the magnetization as a function of the inversion time (TI). The observed signal intensity is given by the following equation: S(t) [ 2exp( TI /T)] () To obtain the transverse relaxation time, T2, values in the brain a series of spin echo images will be acquired by stepping through a range of TEs. The image intensity is given by the following equation: S(t) exp( TE /T 2) (2) T2star (T2*) is the time constant that describes the decay of the transverse magnetization including local magnetic field inhomogeneities. T2* is an important relaxation time for functional studies because it is related to the amount of deoxygenated hemoglobin present in the brain. To determine the T2* values in gray, white matter and CSF, we will be using a gradient echo sequence and the signal is described by: S(t) exp( TE / T 2*) (3)

10 Experiments In this exercise we will acquire and evaluate human data in order to characterize T, T2 and T2* values in brain tissue compartments. Various sequences will be used with the imaging parameters manipulated so that they provide the appropriate contrast weightings. T measurements Acqusition: Total scan time: 5m 38sec A. Inversion Recovery Turbo Spin Echo (tse_ir) sequence with TR=3000ms, FOV=240x240, matrix=96x28, repeated for multiple inversion times TI=200, 300, 400, 500, 600, 700, 800, 000, 300, 500, 700, 2000, 2200 ms.. Load all the images on the viewer. 2. Draw ROIs in gray matter areas. 3. Record the mean signal values in Table for each inversion time and ROI. 4. Repeat steps -3 for regions of white matter and CSF. Lab Question 6 : Draw the measured signal as a function of inversion time for each of the tissue compartments. Calculate the T for gray, white matter and CSF by fitting Eq. () to your data. Hint: you will need to use a nonlinear fitting function such as nlinfit in matlab to accomplish this. T2 measurements Acqusition: Total scan time: 9m 42sec B. SE sequence with TR=4000ms, FOV=220x220, matrix=92x92, 8 echo times in a range between 4.3ms and 4.4ms and inter-echo spacing 4.3 ms. 5. Load all the images on the viewer. 6. Draw ROIs in gray matter areas. 7. Record the mean signal values in Table 2 for each echo time and ROI. 8. Repeat steps -3 for regions of white matter and CSF. Lab Question 7 : Draw the measured signal as a function of echo time for each of the tissue compartments. Calculate the T2 for gray, white matter and CSF by fitting Eq. (2) to your data. T2* measurements Acqusition: Total scan time: m 20sec C. 2D FLASH sequence with TR=00msec, FOV=256x256, matrix=28x28, sequence was repeated for multiple echo times TE=6, 0, 5, 20, 30, 50, 70, 90 msec.

11 9. Load all the images on the viewer. 0. Draw ROIs in gray matter areas.. Record the mean signal values in Table 3 for each echo time and ROI. 2. Repeat steps -3 for regions of white matter and CSF. Lab Question 8 : Draw the measured signal as a function of echo time for each of the tissue compartments. Calculate the T2* of gray and white matter and CSF by fitting Eq. (3) to your data. Lab Question 9 : According to your measurements how do T2 and T2* compare for the same tissue compartment? Did you expect these findings? Explain why. Table T Measurements TI (ms) Gray White CSF Matter Matter

12 Table 2 T2 Measurements TE (ms) Gray White CSF Matter Matter Table 3 T2* Measurements TE (ms) Gray White CSF Matter Matter

13 3. Distortion in EPI due to B0 Inhomogeneity Background The echo-planar image (EPI) is distorted due to local field gradients present in the head during imaging. Since the acquisition of k-space data is very asymmetric in EPI, with the readout direction points (kx) collected quickly and the phase encode steps (ky) relatively slowly, the distortion is essentially entirely in the y direction (phase encode direction). The relative distortion between the two directions can be estimated from the relative speed of the acquisition. In kx, at a typical readout BW (say 2.2kHz/pixel or 40kHz in frequency across the 64 pixel image), the dwell time (time between kx samples) is 7.μs. The ky sampling rate is slower because of the zig-zag trajectory, and is about 64 times slower. This gives a sampling spacing of 64 * 7.μs = 0.45ms. We call this.the "echospacing", (esp), of the sequence, the time between gradient echos (acquisition of the kx=0 point). The effective "bandwidth" in the phase encode direction is therefore across the esp image or /64*esp = 34.7Hz/pixel. Therefore a B0 shift which induces a 34.7Hz frequency shift will induce a shift in this region of pixel. The frequency shifts in the "bad susceptibility" regions of the brain are easily in the 00Hz range. Therefore, the principle metric of how distorted the EPI sequence will be is its effective echo-spacing. For conventional echo-planar imaging, this is basically limited by the gradient slew rate and amplitude. Using parallel acceleration such as SENSE or GRAPPA effectively decreases the esp by a factor of the acceleration factor, (i.e. 2 or 3 fold).

14 Experiments In this lab we will acquire images with different effective esp and compare the distortion in the brain. Acquisition : ) epi_sp460: voxel size = 3. x 3. x 3., Matrix size = 64x64, BW=2520Hz/px, ESP=0.46ms, effective esp =0.46ms 2) epi_sp470_grappa: voxel size = 3. x 3. x 3, Matrix size = 64x64, BW=2520Hz/px, ESP=0.46ms, effective esp =0.46ms 3) epi_sp730: voxel size = 3. x 3. x 3, Matrix size = 64x64, BW=502Hz/px, ESP=0.73ms, effective esp =0.73ms Lab Question 0: Estimate the distortion in frontal lobes by measuring distance on the scanner to some reference feature. Plot the distortion in mm as a function of effective esp.

15 /+ SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_5x5x5_signal TA: :44 PAT: Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 0 [%] R3.0 A0.3 H27. [mm] T > C-2.5 Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 240 [mm] FoV phase 00.0 [%] Slice thickness 5 [mm] TR 2000 [ms] TE 32 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle 90 [deg] Reconstruction Magnitude Fat suppr. Fat sat. Measurements 50 Delay in TR 0 [ms] Multiple series Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation [%] PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Shim mode Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] Adjust volume Standard [V] R3.0 A0.3 H27. [mm] T > C [deg] Physio st Signal/Mode 240 [mm] 240 [mm] 45 [mm] BOLD t-test 0 Threshold 4.00 Window Growing Starting ignore meas 0 Paradigm size Meas Ignore Motion correction 0 Spatial filter 0 Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing Long term 3472 [Hz/Px] 0.36 [ms] EPI factor RF pulse type Gradient mode 48 Normal Fast Dummy Scans 2

16 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_3x3x3_signal TA: :44 PAT: Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 67 [%] R3.0 A0.3 H27. [mm] T > C-2.5 Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 92 [mm] FoV phase 00.0 [%] Slice thickness 3 [mm] TR 2000 [ms] TE 32 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle 90 [deg] Reconstruction Magnitude Fat suppr. Fat sat. Measurements 50 Delay in TR 0 [ms] Multiple series Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation [%] PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Physio st Signal/Mode 92 [mm] 92 [mm] 44 [mm] BOLD t-test 0 Threshold 4.00 Window Growing Starting ignore meas 0 Paradigm size Meas Ignore Motion correction 0 Spatial filter 0 Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing Long term 2520 [Hz/Px] 0.47 [ms] EPI factor 64 RF pulse type Normal Gradient mode Fast Dummy Scans 2 Shim mode Standard Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] [V] Adjust volume R3.0 A0.3 H27. [mm] T > C [deg] 2/+

17 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_.5x.5x.5_signal TA: :44 PAT: Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 233 [%] R3.0 A0.3 H27. [mm] T > C-2.5 Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 92 [mm] FoV phase 00.0 [%] Slice thickness.5 [mm] TR 2000 [ms] TE 30 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle 90 [deg] Reconstruction Magnitude Fat suppr. Fat sat. Measurements 50 Delay in TR 0 [ms] Multiple series Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation [%] 5/8 PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Shim mode Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] Adjust volume Standard [V] R3.0 A0.3 H27. [mm] T > C [deg] Physio st Signal/Mode 92 [mm] 92 [mm] 42 [mm] BOLD t-test 0 Threshold 4.00 Window Growing Starting ignore meas 0 Paradigm size Meas Ignore Motion correction 0 Spatial filter 0 Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing Long term 502 [Hz/Px] 0.75 [ms] EPI factor RF pulse type Gradient mode 28 Normal Fast 3/- Dummy Scans 2

18 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\se_mc TA: 9:42 Voxel size: [mm] Rel. SNR:.00 SIEMENS: se_mc Routine Slice group Slices Dist. factor 00 [%] R0.7 A0.8 H2.2 [mm] T > C-0.3 Phase enc. dir. 0 [deg] Phase oversampling 0 [%] FoV read 220 [mm] FoV phase 00.0 [%] Slice thickness 5 [mm] TR 4000 [ms] TE[] 4.3 [ms] TE[2] 28.6 [ms] TE[3] 42.9 [ms] TE[4] 57.2 [ms] TE[5] 7.5 [ms] TE[6] 85.8 [ms] TE[7] 00. [ms] TE[8] 4.4 [ms] Averages Concatenations Filter Raw filter Coil elements HEA;HEP Contrast MTC Magn. preparation Flip angle 80 [deg] Reconstruction Magnitude Fat suppr. Fat sat. mode Strong Water suppr. Measurements Base resolution 92 Phase resolution 00 [%] Phase partial Fourier 6/8 Filter Raw filter Intensity Weak Slope 25 Filter 2 Large FoV Filter 3 Prescan Normalize Filter 4 Normalize Filter 5 Elliptical filter Interpolation Geometry Multi-slice mode Series Special sat. System Body HEP HEA Save uncombined Scan at current TP MSMA Sagittal Coronal Matrix Coil Mode H 0 [mm] S - C - T Auto (CP) Shim mode Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] Adjust volume Physio st Signal/Mode Tune up [V] Isocenter 0 [deg] 350 [mm] 263 [mm] 350 [mm] Dark blood Inline Subtract 0 Std-Dev-Sag 0 Std-Dev-Cor 0 Std-Dev-Tra 0 Std-Dev-Time 0 MIP-Sag 0 MIP-Cor 0 MIP-Tra 0 MIP-Time 0 Save original images Sequence Introduction Averaging mode Contrasts Bandwidth Allowed delay /- RF pulse type Gradient mode Short term [Hz/Px] 0 [s] Normal Fast

19 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\FLASH_T2star_TE6 + TA: 9.6 [s] Voxel size: [mm] Rel. SNR:.00 USER: FLASH Routine Slice group Slices Dist. factor Phase enc. dir. Phase oversampling FoV read FoV phase Slice thickness TR TE Averages Concatenations Filter Coil elements Contrast MTC Flip angle 5 [deg] Reconstruction Magnitude Fat suppr. Water suppr. Measurements Base resolution 28 Phase resolution 00 [%] Phase partial Fourier 6/8 Filter Raw filter Intensity Weak Slope 25 Filter 2 Large FoV Filter 3 Prescan Normalize Filter 4 Normalize Filter 5 Elliptical filter Interpolation Geometry Multi-slice mode Series 20 [%] R0.7 A0.8 H2.2 [mm] T > C [deg] 0 [%] 256 [mm] 00.0 [%] 5 [mm] 00 [ms] 6 [ms] Raw filter HEA;HEP Sequential Special sat. System Body HEP HEA Save uncombined Scan at current TP MSMA Sagittal Coronal Matrix Coil Mode H 0 [mm] S - C - T Auto (CP) Shim mode Tune up Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] Adjust volume Physio st Signal/Mode Inline Subtract 0 Std-Dev-Sag 0 Std-Dev-Cor 0 Std-Dev-Tra 0 Std-Dev-Time 0 MIP-Sag 0 MIP-Cor 0 MIP-Tra 0 MIP-Time 0 Save original images Sequence Introduction Dimension Averaging mode Contrasts Bandwidth [V] Isocenter 0 [deg] 350 [mm] 263 [mm] 350 [mm] 2D Long term 300 [Hz/Px] Gradient mode RF spoiling Fast line ICE Selection box Spoil me! Test Time darray [] darray [2] darray [3] Second Choice 0 ms [UnitArr] 2 [UnitArr] 22 [UnitArr] /-

20 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\tse_IR TA: 0:26 PAT: Voxel size: [mm] Rel. SNR:.00 SIEMENS: tse Routine Slice group Slices Dist. factor 50 [%] R0.7 A0.8 H2.2 [mm] T > C-0.3 Phase enc. dir. 90 [deg] Phase oversampling 0 [%] FoV read 240 [mm] FoV phase 75.0 [%] Slice thickness 5 [mm] TR 3000 [ms] TE 2 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Magn. preparation Slice-sel. IR TI 000 [ms] Flip angle 80 [deg] Reconstruction Real Fat suppr. Fat sat. mode Strong Water suppr. Measurements Base resolution Phase resolution Phase partial Fourier Filter Raw filter Filter 2 Large FoV Filter 3 Prescan Normalize Filter 4 Normalize Filter 5 Elliptical filter Interpolation [%] PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Save uncombined Scan at current TP MSMA Sagittal H 0 [mm] S - C - T Coronal Shim mode Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] Adjust volume Physio st Signal/Mode Tune up [V] Isocenter 0 [deg] 350 [mm] 263 [mm] 350 [mm] Dark blood Resp. control Inline Subtract 0 Std-Dev-Sag 0 Std-Dev-Cor 0 Std-Dev-Tra 0 Std-Dev-Time 0 MIP-Sag 0 MIP-Cor 0 MIP-Tra 0 MIP-Time 0 Save original images Sequence Introduction Dimension Compensate T2 decay Averaging mode Contrasts Bandwidth Flow comp. Allowed delay Echo spacing 2D Short term 30 [Hz/Px] No 0 [s].7 [ms] Turbo factor RF pulse type Gradient mode Hyperecho 5 Normal Fast /-

21 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_esp730 TA: 2.0 [s] PAT: Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 67 [%] R0.3 A.0 F2. [mm] T > C-0.4 Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 200 [mm] FoV phase 00.0 [%] Slice thickness 3 [mm] TR 2000 [ms] TE 30 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle Reconstruction Fat suppr. Measurements Delay in TR Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation 90 [deg] Magnitude Fat sat. 0 [ms] [%] PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Physio st Signal/Mode 200 [mm] 44 [mm] Shim mode Standard Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] [V] Adjust volume R0.3 A.0 F2. [mm] T > C [deg] 200 [mm] /+ BOLD t-test 0 Threshold 4.00 Window Growing Starting ignore meas 0 Paradigm size Meas Ignore Motion correction 0 Spatial filter 0 Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing EPI factor RF pulse type Gradient mode Dummy Scans 0 Long term 502 [Hz/Px] 0.73 [ms] 64 Normal Fast

22 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_esp460 TA: 2.0 [s] PAT: Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 67 [%] R0.3 A.0 F2. [mm] T > C-0.4 Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 200 [mm] FoV phase 00.0 [%] Slice thickness 3. [mm] TR 2000 [ms] TE 30 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle Reconstruction Fat suppr. Measurements Delay in TR Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation 90 [deg] Magnitude Fat sat. 0 [ms] [%] PAT mode Matrix Coil Mode Geometry Multi-slice mode Series Auto (CP) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Physio st Signal/Mode 200 [mm] 45 [mm] Shim mode Standard Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] [V] Adjust volume R0.3 A.0 F2. [mm] T > C [deg] 200 [mm] 2/+ BOLD t-test 0 Threshold 4.00 Window Growing Starting ignore meas 0 Paradigm size Meas Ignore Motion correction 0 Spatial filter 0 Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing EPI factor RF pulse type Gradient mode Dummy Scans 0 Long term 2520 [Hz/Px] 0.46 [ms] 64 Normal Fast

23 SIEMENS MAGNETOM TrioTim syngo MR 2006T \\USER\INVESTIGATORS\HST_583\PhysicsClass\epi_esp460_grappa2 TA: 8.0 [s] PAT: 2 Voxel size: [mm] Rel. SNR:.00 USER: benner\ep2d_bold_mgh_pro_tb Routine Slice group Slices 9 Dist. factor 67 [%] R0.3 A23.9 H8. [mm] Phase enc. dir [deg] Phase oversampling 0 [%] FoV read 92 [mm] FoV phase 00.0 [%] Slice thickness 3 [mm] TR 2000 [ms] TE 30 [ms] Averages Concatenations Filter Coil elements HEA;HEP Contrast MTC Flip angle Reconstruction Fat suppr. Measurements Delay in TR Base resolution Phase resolution Phase partial Fourier Filter Raw filter Interpolation 90 [deg] Magnitude Fat sat. 0 [ms] [%] PAT mode Accel. factor PE Ref. lines PE Matrix Coil Mode Geometry Multi-slice mode Series GRAPPA 2 32 Auto (Triple) Special sat. System Body HEP HEA Scan at current TP MSMA Sagittal Coronal H 0 [mm] S - C - T Physio st Signal/Mode BOLD t-test Threshold Window Starting ignore meas Paradigm size Meas Motion correction Spatial filter Sequence Introduction Averaging mode Bandwidth Free echo spacing Echo spacing [deg] 92 [mm] 92 [mm] 44 [mm] EPI factor RF pulse type Gradient mode Dummy Scans 0 Shim mode Standard Adjust with body coil Confirm freq. adjustment Assume Silicone Ref. amplitude [H] [V] Adjust volume R0.3 A23.9 H8. [mm] 3/ Growing 0 Ignore 0 0 Long term 2520 [Hz/Px] 0.48 [ms] 64 Normal Fast

24