NA-MIC National Alliance for Medical Image Computing fMRI Data Analysis in Slicer (short dataset tutorial) Steve Pieper Haiying Liu Wendy Plesniak Sandy Wells Cindy Wible Carsten Richter

National Alliance for Medical Image Computing Acknowledgments National Alliance for Medical Image Computing NIH U54EB Neuroimage Analysis Center NIH P41RR FIRST Biomedical Informatics Research Network NCRR/NIH 5 MOI RR Harvard Center for Neurodegeneration and Repair Data was acquired on a Philips Achieva 3T scanner. Courtesy of the Brain Imaging Laboratory, Dartmouth Medical School, a participating NAMIC site.

National Alliance for Medical Image Computing Disclaimer It is the responsibility of the user of 3DSlicer to comply with both the terms of the license and with the applicable laws, regulations and rules.

National Alliance for Medical Image Computing Goal of the tutorial Guiding you step by step through the process of using the fMRIEngine to analyze fMRI data and visualize results within Slicer.

National Alliance for Medical Image Computing Prerequisites This tutorial assumes that you have already completed Slicer Basics: Loading and Viewing Data (Slicer Training 1) Saving Data ( Slicer Training 7) Supporting material:

National Alliance for Medical Image Computing Module current status The fMRIEngine is: A package for analyzing and visualizing brain activations in Slicer’s multi-modality environment. a developing framework for a suite of activation detection algorithms and inference engines; currently it provides a basic linear modeling detector only. a prototype under development; it does not yet have full save/load functionality or full compatibility with other fMRI packages.

National Alliance for Medical Image Computing Computer Resources The short-dataset tutorial teaches you how the module works; it uses a truncated fMRI timeseries to get users quickly familiar with the interface and workflow ( we recommend at least 1GB RAM). The full-dataset tutorial guides you through a full fMRI analysis. For this part, your computer must have adequate processing speed and RAM (we recommend at least 3GB). Please use the appropriate tutorial and dataset for you.

National Alliance for Medical Image Computing Tutorial datasets Download: fMRIEngineTutorial.zip Short dataset: functional scans (file format Analyze): functional01.hdr, functional01.img … functional90.hdr, functional90.img structural scan (file format NifTI): anatomical3T.hdr, anatomical3T.img Full dataset: functional scans (file format Analyze): functional01.hdr, functional01.img … functional90.hdr, functional90.img structural scan (file format NifTI): anatomical3T.hdr, anatomical3T.img

National Alliance for Medical Image Computing Tutorial datasets What loaded data looks like: Functional scans…………. (functional01.img) Structural scan..…………. (anatomical3T.img)

National Alliance for Medical Image Computing Data description Functional (EPI): 45 slices, 90 scans; 3.75 x 3.75 x 3.75 mm 3 voxels (64 x 64 matrix with 240mm FOV). Transverse acquisition. Preprocessing: realigned, motion corrected, normalized to MNI, smoothed. Structural (MPRAGE): 170 sagittal slices; 1x1x1.2 mm 3 voxels; normalized to MNI.

National Alliance for Medical Image Computing Paradigm Finger sequencing fMRI task Blocked motor paradigm Subject alternates simple left and right thumb to finger sequencing (always same thumb to 1,2,3,4 finger order within block). Three cycles each of rest, right hand and left hand. TR = 2 seconds Durations: 10 TRs in all epochs Onsets (in TRs): Rest: Right : Left :

National Alliance for Medical Image Computing Slicer’s fMRIEngine

National Alliance for Medical Image Computing fMRIEngine workflow Load preprocessed functional data Describe paradigm and stimulus schedule Specify linear modeling & estimate model parameters Define contrasts and compute parametric map Statistical inference inspect data & combine with other analyses

National Alliance for Medical Image Computing Overview Part 1: Loading and Previewing Data Part 2: Describing stimulus schedule Part 3: Linear modeling & estimation Part 4: Contrasts & computing SPMs Part 5: Inference & inspection

National Alliance for Medical Image Computing Load Image Sequence 1: Pick Sequence->Load tab 7: Click Apply 2: Click File Browse 6: Input a sequence name 5: Input a file filter 4: Click Load Multiple Files 3: Pick any file in the functional directory

National Alliance for Medical Image Computing Set Image Display 11: operations will apply to currently viewed image in the sequence only… 8: Go to Volumes->Display 9: Adjust Win and Lev to get best display of image data 10: Adjust Lo threshold to mask out non-brain

National Alliance for Medical Image Computing Select Image Sequence 13: Pick Select Tab 14: Specify number of runs = 1 15: Click Add to assign sequence to run 1 16: Click Set Window/Level/Thresholds to apply to all volumes in the sequence 12: Go back to fMRIEngine Visually inspect sequence

National Alliance for Medical Image Computing Overview Part 1: Loading and Previewing Data Part 2: Describing stimulus schedule Part 3: Linear modeling & estimation Part 4: Contrasts & computing SPMs Part 5: Inference & inspection

National Alliance for Medical Image Computing Stimulus schedule 16: Pick Set Up Tab and choose the Linear Modeling detector 17: Choose Paradigm Design 18: Enter schedule for first condition (name = right) 19: Click OK to add this condition to the list of defined conditions } detection technique linear modeling step

National Alliance for Medical Image Computing Stimulus schedule 20: Enter schedule for next condition (Name = left) 21: Click OK to add this condition to the list too… }

National Alliance for Medical Image Computing Stimulus schedule 22: To change the schedule for any condition: select condition from box, click Edit, make changes, and click OK again to re-assign it. 23: To delete any defined condition: select it and click Delete Making modifications… After clicking OK, each specified condition for Run 1 is listed as shown below, labeled “r1:condition_name”

National Alliance for Medical Image Computing Overview Part 1: Loading and Previewing Data Part 2: Describing stimulus schedule Part 3: Linear modeling & estimation Part 4: Contrasts & computing SPMs Part 5: Inference & inspection

National Alliance for Medical Image Computing Model Signal 24: Choose the Signal Modeling step 25: Click to model both conditions identically 28: model drift with Discrete Cosine set 27: convolve with HRF 30: Add to Model 26: choose boxcar waveform 29: Choose to use default cutoff

National Alliance for Medical Image Computing View Design Matrix 31: Click View Design 32: The model design matrix appears in a separate window (this may take some time)

National Alliance for Medical Image Computing Estimating model parameters 33: Choose the estimation step 34: Select run1 and click Fit Model

National Alliance for Medical Image Computing Overview Part 1: Loading and Previewing Data Part 2: Describing stimulus schedule Part 3: Linear modeling & estimation Part 4: Contrasts & computing SPMs Part 5: Inference & inspection

National Alliance for Medical Image Computing Specify Contrasts 35: Choose Contrasts step 36: Specify Name (c1) 37: Name activation volume 38: Specify contrast vector (1 –1 0 0) 39: Click OK to add this contrast to a list of defined contrasts 40: Contrast name appears here

National Alliance for Medical Image Computing Check contrasts & model 41: Click View Design 42: Verify contrast is correct

National Alliance for Medical Image Computing Perform activation detection 43: Select Detect tab 44: Select contrast to compute for 45: Click Compute

National Alliance for Medical Image Computing Overview Part 1: Loading and Previewing Data Part 2: Describing stimulus schedule Part 3: Linear modeling & estimation Part 4: Contrasts & computing SPMs Part 5: Inference & inspection

National Alliance for Medical Image Computing Threshold, visualize, inspect 46: Pick View Tab 47: Select activation volume 48: Click Select

National Alliance for Medical Image Computing Threshold, visualize, inspect 49: Pick Threshold Tab 50: Select uncorrected p-value 51: Specify p-value threshold = Activations appear in the foreground

National Alliance for Medical Image Computing Threshold, visualize, inspect Display anatomical image in background Color map code: Strong negative activation Below p-value threshold (not rendered) Weak negative activation Weak positive activation Strong positive activation

National Alliance for Medical Image Computing Threshold, visualize, inspect 52: Pick Plot tab 53: Select condition = right 54: Select Timecourse plot option Plotting…

National Alliance for Medical Image Computing Threshold, visualize, inspect 55: Mouse over labeled voxel in Slice Window and click… view voxel’s timecourse plotted with the modeled condition

National Alliance for Medical Image Computing Threshold, visualize, inspect 56: Select Peristimulus histogram option and click a voxel labeled as active 57: Click any curve on the peristimulus plot for details

National Alliance for Medical Image Computing Threshold, visualize, inspect 58: Select ROI tab 59: Choose New from label map pulldown menu… Activation-based Regions of Interest…

National Alliance for Medical Image Computing Threshold, visualize, inspect 60: Click Create label map… 61: Wait short while while activation “blobs” are labeled 62: Default blob view: Label map shown in FG; activation map shown in BG.

National Alliance for Medical Image Computing Threshold, visualize, inspect 63: Select Stats tab 64: Select one or multiple regions to include in analysis by clicking in Slice Window. Selected regions display green. 65: (clear all selections and start over if you make a mistake)

National Alliance for Medical Image Computing Threshold, visualize, inspect 66: Click Show stats View region statistics

National Alliance for Medical Image Computing Threshold, visualize, inspect 67: Select Timecourse plot option 68: Click Plot time series for this region View region statistics

National Alliance for Medical Image Computing Threshold, visualize, inspect 69: Select Peristimulus histogram option 70: Click Plot time series for this region View region statistics

National Alliance for Medical Image Computing Threshold, visualize, inspect 72: Fade in the activation volume a little more for a good view of combined data 71: Clear selections. Then display anatomical image in BG and activation in FG.

National Alliance for Medical Image Computing Threshold, visualize, inspect Adjusting colormaps for visualization… 73: Navigate to Volumes->Display 74: Set the Active Volume to be the activation volume 75: Adjust Window and Level to shift the color map hue range Recall color map code: Strong negative activation Below p-value threshold (not rendered) Weak negative activation Weak positive activation Strong positive activation

National Alliance for Medical Image Computing Threshold, visualize, inspect To view only positive activations… 76: Choose Palette to be fMRIPosActive 75: Adjust Window and Level to shift the color map hue range Color map code: Below p-value threshold (not rendered) Weak positive activation Strong positive activation

National Alliance for Medical Image Computing Features to come… Features that will be developed within the fMRIEngine: Load/Save functionality for modeling and analysis Ability to apply prior models Ability to perform other statistical tests Other techniques for performing activation detection Advanced visualization …