Category-specific Associative Inference in Memory Dataset

# Category-specific Associative Inference in Memory Dataset

This dataset contains fMRI, behavioural and eye-tracking data on a experiment testing the involvement of Medial Temporal Lobe (MTL) subregions in category-specific associative inference in memory. The dataset follows the Brain Imaging Data Structure ([BIDS](https://bids.neuroimaging.io/)) format.

## Dataset Structure

```
├── dataset_description.json
├── participants.json
├── participants.tsv
├── README.md
├── derivatives/
│ ├── AIMDeconvolveOutput
│ │ └── sub-[subjectID]/
│ ├── LocalDeconvolveOutput
│ │ └── sub-[subjectID]/
│ └── ROIs
│ └── sub-[subjectID]/
└── sub-[subjectID]/
├── anat/
│ ├── sub-[subjectID]_T1w.nii.gz
│ └── sub-[subjectID]_T2w.nii.gz
└── func/
├── sub-[subjectID]_task-aim_run-[01-08]_bold.nii.gz
├── sub-[subjectID]_task-aim_run-[01-08]_bold.json
├── sub-[subjectID]_task-local_bold.nii.gz
├── sub-[subjectID]_task-local_bold.json
├── sub-[subjectID]_task-aim_events.tsv
├── sub-[subjectID]_task-aim_events.json
├── sub-[subjectID]_task-local_events.tsv
├── sub-[subjectID]_task-local_events.json
├── sub-[subjectID]_task-aim_eyetracking.tsv
└── sub-[subjectID]_task-aim_eyetracking.json
```

## Dataset Contents

### Anatomical Data
- T1-weighted structural scans (defaced)
- T2-weighted structural scans (slab collected perpendicular to the long-axis of the hippocampus)

### Raw Functional Data
1. Associative Inference Memory (AIM) Task
- 8 runs of BOLD data
- Event files containing task timing and conditions
- Eye tracking data
- Associated JSON metadata files

2. Local Task
- Single run of BOLD data
- Event files containing task timing and conditions
- Associated JSON metadata files

### Derivatives Dtaa
1. AIM Deconvolve Output (`AIMDeconvolveOutput`)
- Whole-brain maps containing results from the GLM analysis: sub-[subjectID]_StudyIndirectSubMemTestAcc_StrictAcc.nii.gz
- Design matrix generated by 3dDeconvolve in AFNI (v21.2.10): sub-[subjectID]_StudyIndirectSubMemTestAcc_StrictAcc_glm_design1.1D
- 3dREMLfit command generated by 3dDeconvolve in AFNI (v21.2.10): sub-[subjectID]_StudyIndirectSubMemTestAcc_StrictAcc+tlrc.REML_cmd
2. Localiser Deconvolve Output (`LocalDeconvolveOutput`)
- Whole-brain maps containing results from the GLM analysis: sub-[subjectID]_LocCat.nii.gz
- Design matrix generated by 3dDeconvolve in AFNI (v21.2.10): sub-[subjectID]_LocCat_glm_design1.1D
- 3dREMLfit command generated by 3dDeconvolve in AFNI (v21.2.10): sub-[subjectID]_LocCat+tlrc.REML_cmd
3. Regions of Interest (`ROIs`)
- Manually segmented MTL subregion ROIs in T1 space: sub-1004_space-T1w_desc-HandSeg_mask.nii.gz</br>The manual segmentation was conducted on raw hi-resolution T2w images. These were transformed to T1w space using the following steps:
<ol type="a"><li>The raw T2w image was aligned to the T1w image using <code>antsRegistrationSynQuick</code> in ANTS.</li>
<li>The transform computed in the above step was applied to the manually segmented ROIs in the raw T2w space using <code>antsApplyTransform.sh</code> in ANTS.</li></ol>
- Manually segmented MTL subregion ROIs resampled to BOLD space: sub-1004_space-T1w_desc-HandSeg_mask_Resampled.nii.gz</br>Because of differences in obliquity of the BOLD and T1 scans following preprocessing in fMRIprep, the following steps were undertaken to resample the ROIs from T1 space to BOLD space (see [here](https://discuss.afni.nimh.nih.gov/t/3dresample-on-fmriprep-output-moves-dset-around/7669/4) for more details):
<ol type="a"><li>The difference in obliquity between the preprocessed T1w image and the preprocessed BOLD image was computed using <code>3dWarp</code> in AFNI.</li>
<li>The obliquity difference was applied to the ROIs in T1w space using <code>3dAllineate</code> in AFNI.</li></ol>

### Metadata Files
- `dataset_description.json`: Dataset-level metadata
- `participants.tsv`: Participant information
- `participants.json`: Description of participant-level variables

## Notes
- All anatomical images have been defaced for privacy
- The dataset includes 41 participants who completed the AIM scan, and 38 participants who completed the funcational localiser scan (sub-1003 through sub-1041)
- Eye tracking data is available for the AIM task
- Scripts used to analyse this dataset are available on [Github](https://https://github.com/mrinmayik/CategorySpecificAssociativeInference/)