Data from: ASHS-OAP atlas for automatic entorhinal cortex segmentation
Data files
Feb 27, 2024 version files 4.01 GB
-
ashs_oap_atlas.zip
4.01 GB
-
README.md
4.86 KB
Abstract
Early stages of Alzheimer’s disease (AD) are associated with volume reductions in specific subregions of the medial temporal lobe (MTL). Using a manual segmentation method—the Olsen-Amaral-Palombo (OAP) protocol— previous work in healthy older adults showed that reductions in grey matter volumes in MTL subregions were associated with lower scores on the Montreal Cognitive Assessment (MoCA), suggesting atrophy may occur prior to diagnosis of mild cognitive impairment, a condition that often progresses to AD. However, current manual segmentation methods are labour intensive and time consuming. Here, we examined the utility of Automatic Segmentation of Hippocampal Subfields (ASHS) to detect volumetric differences in MTL subregions of healthy older adults who varied in cognitive status as determined by the MoCA. We trained ASHS on the OAP protocol to create the ASHS-OAP atlas, and then examined how well automated segmentation replicated the ground truth of manual segmentation. Volumetric measures obtained from the ASHS-OAP atlas were also contrasted against those from the ASHS-PMC atlas, a widely used atlas provided by the ASHS team. Volumetrics from the ASHS-OAP atlas aligned well with those from manual segmentation, suggesting ASHS-OAP is a viable alternative to current manual segmentation methods. In addition, while some subtle differences were observed, results from the ASHS-PMC and ASHS-OAP atlases aligned well with each other overall. Our findings highlight the utility of automated segmentation methods but still underscore the need for a unified and harmonized MTL segmentation atlas.
README: ASHS-OAP atlas for automatic entorhinal cortex segmentation
ASHS is an open-source tool designed for the Automated Segmentation of Hippocampal Subfields from MRI images of the brain. Several atlases for the hippocampus are provided with ASHS. While it was developed for the segmentation of hippocampal subfields, the tools provided with ASHS can potentially be used for multi-atlas segmentation of any region if an appropriate set of expert-derived atlas segmentations is provided. Here, we extend ASHS by providing an atlas to segment subregions of the adjacent entorhinal cortex.
ASHS allows users to run the segmentation algorithm with a custom atlas (via the -a ATLASDIR parameter). Here, we provide all of the files required for ASHS to operate on the entorhial cortex. These were derived from expert manual segmentation of the regions and by using ASHS to create and cross-validate all the required files (see manuscript for details). Users must only install these sets of files in a folder and direct ASHS to use this folder as the template to perform automatic entorhinal cortex segmentation.
Briefly, the provided atlas files that ASHS uses include:
adaboost: This folder contains files that have information produced by the machine learning algorithm (adaboost)that ASHS uses for segmentation.ashs_atlas_vars.sh: This file contains information about the ASHS version and date.ashs_system_config.sh: This file contains default settings for ASHS.ashs_user_config.sh: This file contains information for the configuration of ASHS.snap: This folder contains a file that is used to specify the names of the anatomical labels used in the protocol as well as the color/value that should be used when generating figures. This file is read by ITK-SNAP.template: This folder has files that contain template and reference MRI scans and labels/masks for ASHS to use for segmentation. Note that the main template structuraltemplate.nii.gz(and thetemplate_bet.nii.gzbrain-extracted version) is a central-tendency model abstracted from the 40 participants. The variousrefspaceimages are crops (and masks thereof) used by ASHS. Each of these files was created by ASHS during its template creation process and are required for its multi-atlas segmentation of a new participant.train: This folder has 40 sub-folders (one for each participant) that contain the MRI scans (T1 and T2), manual segmentations, and processed images (co-registered and cropped). Each folder represents one of the 40 participants that ASHS was trained on to create this atlas. During the ASHS atlas creation, each training subject's images are cropped ("chunk") and registered to the template using both affine (.mat) and non-linear ("warp") registration. Each of these files was generated by ASHS and is required for its multi-atlas segmentation of a new participant.
Description of the data and file structure
This is an atlas set that can be used with Automatic Segmentation of Hippocampal Subfields (ASHS) software to perform automatic segmentation of medial temporal lobe subregions.
This atlas set was constructed from MRI scans and manual segmentations shared by Olsen, Yeung et al. (2017).
The atlas contains segmented scans of 40 healthy older adults acquired at 3 Tesla. The T2-weighted scans are at 0.43x0.43x3mm3 resolution. The
manual segmentations include nine labels per hemisphere. These regions include the anterior head and tail of the hippocampus, three subfields of the hippocampus (CA1, CA2/CA3/DG, subiculum) and four medial temporal lobe cortex subregions (anterolateral entorhinal cortex, posteromedial entorhinal cortex, perirhinal cortex, parahippocampal cortex).
Please see the primary ASHS installation site for instructions on how to download and install ASHS software and atlas package to your local machine.
Please see the primary ASHS documentation for instructions on how to run ASHS using a provided atlas package.
The dataset provided here make up a custom atlas package. Use this by simply specifying the folder you have extracted this to instead of the ASHS-provided atlas when calling the ashs_main script (the -a ATLASDIR parameter) in order to segment new datasets according to the provided package.
Code/Software
In addition to the provided package, please see the primary ASHS installation site on how to download and install ASHS software on your local machine. A working installation of the open-source ASHS is required for this package to be used.
Methods
Forty community-dwelling healthy older adults (as described in Olsen, Yeung et al., 2017; age range = 59-81 years; mean age = 71.4; mean education = 16.3 years; range = 12-23; 30 female) were recruited from participant databases at the Rotman Research Institute and the University of Toronto.
Neuroimaging was completed on a 3-Tesla Siemens Trio scanner using a 12-channel head coil. All participants received a T1-weighted, magnetization-prepared, rapid acquisition with gradient echo image (MP-RAGE) whole-brain anatomical scan (TE/TR=2.63ms/2000ms, 160 axial slices perpendicular to the AC-PC line, voxel size=1x1x1mm, FOV=256mm). The T1-weighted MP-RAGE was used for slice placement during the acquisition of a subsequent high in-plane resolution T2-weighted scan in an oblique-coronal plane, perpendicular to the hippocampal long axis (TE/TR=68/3000ms, 20-28 slices depending on head size, 512x512 acquisition matrix, voxel size=0.43x0.43x3mm, no skip, FOV=220mm). For the T2-weighted scan, the first slice was placed anterior to the collateral sulcus (including the temporal pole where possible) and the last slice was placed posterior to the hippocampal tail to ensure full coverage of the entire hippocampus and MTL cortices for all participants.
A custom atlas was built using the Automatic Segmentation of Hippocampal Subfields (ASHS) software (Yushkevich et al., 2015) following the published procedures (https://sites.google.com/view/ashs-dox/local-ashs/building-an-atlas-for-t2-mri), without the slice heuristics procedure. We generated a custom atlas package (ASHS-OAP) based on the forty healthy older adults’ manual segmentations from Olsen, Yeung et al. (2017). Atlas building in ASHS was performed on a Linux cluster with a Sun Grid Engine. For more details on the OAP segmentation protocol see Olsen et al. (2013), Palombo et al. (2013), and https://github.com/NataliaLadyka/OAPSegmentation/blob/main/Comprehensive%20MTL%20Segmentation%20Protocol%20-%20Updated%20.pdf.
The automated segmentation will result in nine regions of interest per hemisphere. These regions are the anterior head and tail of the hippocampus, three subfields of the hippocampus (CA1, CA2/CA3/DG, subiculum) and four medial temporal lobe cortex subregions (anterolateral entorhinal cortex, posteromedial entorhinal cortex, perirhinal cortex, parahippocampal cortex).