Whole Brain Histogram Analysis for Amyloid PET
A semi‑quantitative analysis method for amyloid PET that does not require normalization
Embedded Files
Tutorial
Tutorial
This tutorial explains how to perform WBHA analysis in practice. While the ANM paper evaluates methods such as HD‑BET and FSL for brain extraction, these methods cannot calculate the top 20% GW‑ratio. In this tutorial, we focus exclusively on the SPM method, which is capable of calculating the top 20% GW‑ratio.
Required Analysis Environment
Required Analysis Environment
- FSL - FMRIB Software Library
https://fsl.fmrib.ox.ac.uk/fsl/docs/
FSL is used to generate mask images from MRI data segmented with SPM. It is also used to extract brain parenchyma from registered PET images.
- SPM12
https://www.fil.ion.ucl.ac.uk/spm/software/spm12/
SPM is used to perform co‑registration and reslicing of PET and structural MRI images, as well as to segment MRI data.
If you are not familiar with installing FSL or SPM12, you can also use a pre‑configured virtual environment.
Lin4Neuro: A virtual environment that bundles the software required for brain extraction
https://www.nemotos.net/?page_id=29 (English)
https://www.nemotos.net/?page_id=161 (日本語)
Lin4Neuro (L4N) is based on Ubuntu with XFCE desktop environment. Lin4Neuro includes many useful neuroimaging software packages which developed by Dr. Kiyotaka Nemoto (https://researchmap.jp/k-nemoto?lang=en). The following FSL and SPM12 are also included.
- MATLAB or MATLAB runtime
https://jp.mathworks.com/products/matlab.html
We recommend using MATLAB itself when processing large amounts of data.
https://jp.mathworks.com/products/compiler/matlab-runtime.html (Version 2025a Recommended)
You will need the MATLAB Runtime to use the pre‑compiled analysis program with a GUI that we have created.
- Our in-house Shell script and MATLAB scripts
●Linux Shell FSL Script for brain extraction, generate Top-20% map, and separate Top-20% map into grat matter and white matter (Step 4 to 8)
https://kakuigaku.info/share/PET_and_MRI_Brain_Extraction_SPM
●MATLAB Script for generate Histogram, Skewness, MMR, and Top 20% GW-ratio (Step9)
https://kakuigaku.info/share/PET_Nifti_to_skewness_MMR_and_gwratio.m
●MATLAB GUI Script for generate Histogram, Skewness, MMR, and Top 20% GW-ratio (Step9')
https://kakuigaku.info/share/WBHA_calculator_ver1.m
●MATLAB GUI Application for generate Histogram, Skewness, MMR, and Top 20% GW-ratio (Step9')
https://kakuigaku.info/share/WBHA_calculator_ver1_for_Windows.zip
https://jp.mathworks.com/products/compiler/matlab-runtime.html (Version 2025a MATLAB runtime Recommended)
Workflow overview
Workflow overview
1. Co-registration and Re-slice (SPM12)
1. Co-registration and Re-slice (SPM12)
PET images were spatially aligned to the structural MRI space through co-registration and reslicing using SPM12
2. Smoothing (SPM12)
2. Smoothing (SPM12)
Spatial smoothing was applied to the co-registered images utilizing a Gaussian kernel with a 5 mm full width at half maximum (FWHM).
3. Segmentation (SPM12)
3. Segmentation (SPM12)
Structural MRI was first processed using tissue segmentation in SPM12 . This generated six tissue probability maps (c1–c6), corresponding to gray matter (GM, c1), white matter (WM, c2), cerebrospinal fluid (CSF, c3), bone (c4), soft tissue (c5), and background (c6).
4. Making mask image (FSL)
4. Making mask image (FSL)
The brain parenchyma mask was constructed by combining the GM (c1) and WM (c2) probability maps via FSL.
5. PET brain extraction (FSL)
5. PET brain extraction (FSL)
The smoothed, resliced, and co-registered PET data were then masked by the resultant brain-extracted MRI template to isolate the brain parenchyma.
6. Top 20% map generate (FSL)
6. Top 20% map generate (FSL)
Top 20% maps are generated from a brain-extracted PET image and FSL's "-thrP 80" option
7. Extract Top 20% map in gray matter (FSL)
7. Extract Top 20% map in gray matter (FSL)
Top 20% maps are subsequently segmented into Gray matter (GM, c1) components using tissue probability maps derived from the SPM processing, with a threshold set at 0.5.
8. Extract Top 20% map in white matter (FSL)
8. Extract Top 20% map in white matter (FSL)
Top 20% maps are subsequently segmented into wray matter (WM, c2) components using tissue probability maps derived from the SPM processing, with a threshold set at 0.5.
9. Run our MATLAB script (MATLAB)
9. Run our MATLAB script (MATLAB)
Our MATLAB script calculates skewness, MMR, and the top 20% GW-ratio.
Ⅰ.Internal MATLAB processing for skewness calculation
Ⅰ.Internal MATLAB processing for skewness calculation
Extract the non‑zero voxel values as a matrix, and then calculate the skewness of this matrix.
Ⅱ.Internal MATLAB processing for MMR calculation
Ⅱ.Internal MATLAB processing for MMR calculation
(A) Data conversion and extraction: Voxel values were extracted and ordered sequentially.
(B) Normalization: Voxel values were normalized to a range of 0–1000.
(C) MMR preparation: Non‑zero normalized voxel values were isolated and rounded to the nearest integer for mode calculation.
(D) MMR determination: The mode and mean were calculated from the prepared non‑zero data, and the MMR was subsequently computed.
MMR = Mode value / Mean value
Ⅲ .Internal MATLAB processing for Top 20% GW-ratio calculation
Ⅲ .Internal MATLAB processing for Top 20% GW-ratio calculation
The resulting voxels in the separated Top 20% GM and WM maps were counted using MATLAB.
The Top 20% GW-ratio was then calculated using the following formula:
Hands‑on analysis
Hands‑on analysis
1. Co-registration and Re-slice (SPM12)
1. Co-registration and Re-slice (SPM12)
Select "Coregister (EST & Res)"
Reference Image = Structural MRI
Source Image = PET image
Run▶
2. Smoothing (SPM12)
2. Smoothing (SPM12)
Select "Smooth"
Change FWHM setting into Gaussian kernel with a 5 mm.
Run▶
3. Segmentation (SPM12)
3. Segmentation (SPM12)
Select "Segment"
Volumes = Structural MRI
Run ▶
4-8 Linux shell script (FSL)
4. Making mask image (FSL), 5. PET brain extraction (FSL), 6. Top 20% map generate (FSL), 7. Extract Top 20% map in gray matter (FSL), 8. Extract Top 20% map in white matter (FSL)4-8 Linux shell script (FSL)
Copy the processing script into the folder containing the files to be processed.
Navigate to the folder containing the files processed by SPM using the "cd" command.
Run our in-house shell script "PET_and_MRI_Brain_Extraction_SPM"
9. Run our MATLAB script (MATLAB)
9. Run our MATLAB script (MATLAB)
Navigate to the folder containing the files processed by SPM and FSL
Copy the processing MATLAB script into the folder containing the files to be processed.
Run▶
9'. Run our MATLAB GUI script or app (MATLAB)
9'. Run our MATLAB GUI script or app (MATLAB)
If you use our GUI script or GUI app (Windows standalone app)
Click "Select"and 1:brain extracted PET image, 2:Top 20% map in GM, and3: Top 20% map in WM
Run▶
10. Result
10. Result
Histogram
CSV file (Contain Skewness, MMR, and Top 20% GW-ratio)
Top 20% map in Gray matter
(Overlay on structural MRI)
Page updated
Google Sites
Report abuse