CAT – A Computational Anatomy Toolbox for the Analysis of Structural MRI Data

  1. Christian Gaser
  2. Robert Dahnke
  3. Paul M Thompson
  4. Florian Kurth
  5. Eileen Luders
  6. Alzheimer’s Disease Neuroimaging Initiative

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

A large range of sophisticated brain image analysis tools have been developed by the neuroscience community, greatly advancing the field of human brain mapping. Here we introduce the Computational Anatomy Toolbox (CAT) – a powerful suite of tools for brain morphometric analyses with an intuitive graphical user interface, but also usable as a shell script. CAT is suitable for beginners, casual users, experts, and developers alike providing a comprehensive set of analysis options, workflows, and integrated pipelines. The available analysis streams – illustrated on an example dataset – allow for voxel-based, surface-based, as well as region-based morphometric analyses. Notably, CAT incorporates multiple quality control options and covers the entire analysis workflow, including the preprocessing of cross-sectional and longitudinal data, statistical analysis, and the visualization of results. The overarching aim of this article is to provide a complete description and evaluation of CAT, while offering a citable standard for the neuroscience community.

Article activity feed

  1. GigaScience

    AbstractA large range of sophisticated brain image analysis tools have been developed by the neuroscience community, greatly advancing the field of human brain mapping. Here we introduce the Computational Anatomy Toolbox (CAT) – a powerful suite of tools for brain morphometric analyses with an intuitive graphical user interface, but also usable as a shell script. CAT is suitable for beginners, casual users, experts, and developers alike providing a comprehensive set of analysis options, workflows, and integrated pipelines. The available analysis streams – illustrated on an example dataset – allow for voxel-based, surface-based, as well as region-based morphometric analyses. Notably, CAT incorporates multiple quality control options and covers the entire analysis workflow, including the preprocessing of cross-sectional and longitudinal data, statistical analysis, and the visualization of results. The overarching aim of this article is to provide a complete description and evaluation of CAT, while offering a citable standard for the neuroscience community.

    A version of this preprint has been published in the Open Access journal GigaScience (see paper https://doi.org/10.1093/gigascience/giae049), where the paper and peer reviews are published openly under a CC-BY 4.0 license. These peer reviews were as follows:

    Reviewer 3: Cyril Pernet

    CAT has been around for a long time and is a well maintained toolbox - the paper describes all the features and additionally provides tests/validations of those features. I have left a few comments on the pdf (uploaded) which I don't see has mandatory and thus 'accepted' the paper (and leave the authors to decide what to do with those comments). It provides a nice reference for the toolbox.

  2. GigaScience

    AbstractA large range of sophisticated brain image analysis tools have been developed by the neuroscience community, greatly advancing the field of human brain mapping. Here we introduce the Computational Anatomy Toolbox (CAT) – a powerful suite of tools for brain morphometric analyses with an intuitive graphical user interface, but also usable as a shell script. CAT is suitable for beginners, casual users, experts, and developers alike providing a comprehensive set of analysis options, workflows, and integrated pipelines. The available analysis streams – illustrated on an example dataset – allow for voxel-based, surface-based, as well as region-based morphometric analyses. Notably, CAT incorporates multiple quality control options and covers the entire analysis workflow, including the preprocessing of cross-sectional and longitudinal data, statistical analysis, and the visualization of results. The overarching aim of this article is to provide a complete description and evaluation of CAT, while offering a citable standard for the neuroscience community.

    A version of this preprint has been published in the Open Access journal GigaScience (see paper https://doi.org/10.1093/gigascience/giae049), where the paper and peer reviews are published openly under a CC-BY 4.0 license. These peer reviews were as follows:

    Reviewer 2: Chris Foulon

    Overall, I think the CAT software provides valuable tools to analyse morphometric differences in the brain and promotes open science. The study shows the software's capabilities rather well. However, I think some clarifications would help the readers understand and evaluate the quality of the methods.

    Comments: Figure 2: Looking at the chart, I have a question regarding the pipeline. Is it required to run the whole pipeline using CAT? Or is it possible to input already registered data to start directly with the VBM analysis or further?

    Voxel-based Processing: The above question is quite important, seeing that the preprocessing uses rather old registration methods. The users might want to use more recent registration methods, especially with clinical populations.

    Spatial Registration and Figure 3: For the registration, how is the registration performing with clinical populations (e.g. stroke patients)? It can be significant for the applicability of the methods with specific disorders.

    Surface Registration and Figure 3: What type of noise is used to evaluate the accuracy? This can be important as not every noise can be modelled easily, and some noises are more or less pronounced depending on the modality.

    Maybe having the letters of the figure panels referred to in the text would help the reader.

    Performance of CAT: Although I see the advantage of using simulated data, I think it would require more explanation. First, what tells the reader the quality of this simulated data, and how does it compare to real data? Second, is it only healthy data? In that case, the accuracy evaluation might not be relevant for the majority of the clinical studies using CAT.

    Longitudinal Processing: Are VBM analyses sensitive enough to capture changes over days? I would be surprised, but I would be interested to see studies doing it (and the readers would also benefit from it, I reckon).

    Mapping onto the Cortical Surface: I am a bit confused about the interest in mapping functional or diffusion parameters to the surface. Do you have examples of articles doing that? It sounds like it would waste a lot of information from these parameters, but I am not familiar with this type of analysis. "Optionally, CAT also allows mapping of voxel values at multiple positions along the surface normal at each node". I do not understand this sentence; I think it should be clarified.

    Example application: Is there a way to come back from the surface space to the volume space to compare the results? For example, VBM and SBM should provide fairly similar results, but comparing them is difficult when they are not in the same space. Additionally, in the end, the surface representation is just that, a representation; most other analyses are still done on the volume space, so it could be helpful to translate the result on the surface back to the volume (if it is not already available).

    Evaluation of CAT12: I was confused with Supplemental Figure 1 as it is not mentioned in the caption that it is the AD data and not the simulated one. Maybe it would help the reader to mention it.

    Regarding the reliability of CAT12, it seems to capture more things, but I struggle to see how we can be sure that this is "better" than other methods; couldn't it be false positives?

    "those achieved based on manual tracing and demonstrated that both approaches produced comparable hippocampal volume." comparable volumes do not really mean the same accuracy; this sentence could be misleading.

    I think the multiple studies show that CAT12 is as valid as any other tool but I am not sure the argument that it is better is as solid. Of course, I understand that there is no ground truth for what a relevant morphological change is for a given disease.

    Methods: Statistical Analysis: Why is the FWER correction used for the voxel-wise statistics (which perform many comparisons) and FDR used on ROI-wise statistics (which perform much fewer comparisons)? I would expect the opposite.

    "The outcomes of the VBM and voxel-based ROI analyses were overlaid onto orthogonal sections of the mean brain created from the entire study sample (n=50); " I don't understand what this refers to.

  3. GigaScience

    AbstractA large range of sophisticated brain image analysis tools have been developed by the neuroscience community, greatly advancing the field of human brain mapping. Here we introduce the Computational Anatomy Toolbox (CAT) – a powerful suite of tools for brain morphometric analyses with an intuitive graphical user interface, but also usable as a shell script. CAT is suitable for beginners, casual users, experts, and developers alike providing a comprehensive set of analysis options, workflows, and integrated pipelines. The available analysis streams – illustrated on an example dataset – allow for voxel-based, surface-based, as well as region-based morphometric analyses. Notably, CAT incorporates multiple quality control options and covers the entire analysis workflow, including the preprocessing of cross-sectional and longitudinal data, statistical analysis, and the visualization of results. The overarching aim of this article is to provide a complete description and evaluation of CAT, while offering a citable standard for the neuroscience community.

    A version of this preprint has been published in the Open Access journal GigaScience (see paper https://doi.org/10.1093/gigascience/giae049), where the paper and peer reviews are published openly under a CC-BY 4.0 license. These peer reviews were as follows:

    Reviewer 1: Chris Armit

    This Technical Note describes the Computational Anatomy Toolbox (CAT) software tool, which includes a Graphical User Interface that can be used for morphometric analysis of Structural MRI data. The CAT software tool is impressive, and enables voxel-based and surface-based morphometric analysis to be accomplished on Structural MRI data, and also voxel-based tissue segmentation and surface mesh generation to be applied to these 3D imaging datasets. The authors helpfully illustrate the utility of the Computational Anatomy Toolbox (CAT) using T1-weighted structural brain images from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database.

    This is an excellent, freely available tool for the Neuroimaging community and the authors are to be commended for developing this impressive software tool.

    Minor comments

    I first attempted to launch the CAT software tool on macOS 14.0 (Sonoma) with Apple M1 chip, and on the command line I received the following message: "spm12" is damaged and can't be opened. You should move it to the Bin.

    I additionally tested the CAT software tool on macOS 12.6 (Monterey) with Intel chip, and I was able to run the CAT software tool on this platform.

    A minor criticism is that the installation instructions in the supporting Readme file for archive [CAT12.9_R2023b_MCR_Mac_arm64.zip], which runs on macOS with Intel chip, only details how to install the SPM (Statistical Parametric Mapping) software tool. The CAT software tool needs to be downloaded separately and then moved into the directory of the SPM toolbox, and these installation instructions are included in the supporting CAT software documentation (https://neuro-jena.github.io/cat12-help/#get_started)

    With the issues I encountered in installation, I invite the authors to list the System Requirements - specifically the Operating Systems that are needed to run the CAT software tool - in the GigaScience manuscript and also in the supporting CAT software documentation.

    In addition, it would be particularly helpful if the instructions on how to install CAT in the context of SPM were included in the supporting Readme files for the Computational Anatomy Toolbox (CAT) zip archives.

  4. Version published to 10.1101/2022.06.11.495736v2 on bioRxiv
  5. Version published to 10.1101/2022.06.11.495736v1 on bioRxiv