Epileptic seizures are generated in cerebral networks that propagate ictal and interictal activity. The structure of cerebral networks underpinning epileptic activity can be inferred from diffusion-weighted MRI (DWI). However, publicly available DWI data in individuals with epilepsy are scarce, and processing is technically challenging due to scan-specific artifacts, limiting research progress. Here, we release raw DWI data from 216 individuals with epilepsy and 98 healthy controls. Subject identifiers align with our previous data release (IDEAS), which includes T1-weighted and FLAIR MRI, surgical details, and long-term seizure outcomes after surgery. Preprocessing reduced distortions and artifacts, while fully processed data include diffusion metric maps in native and template space. We also provide parcellated structural connectomes using multiple atlases and connectivity measures. To illustrate the utility of this IDEAS II data, we replicated ENIGMA consortium findings, observing widespread reductions of fractional anisotropy, particularly ipsilateral to the area of seizure onset. We further demonstrate localised abnormality, and network connectivity using streamline tractography in a patient who subsequently underwent temporal lobe resection. This open dataset offers a comprehensive resource to advance research on structural connectivity and surgical outcomes in epilepsy.
Data sharing initiatives such as the human connectome project, ADNI, and ABIDE have facilitated major progress in their respective fields [1][2][3][4][5] . In epilepsy neuroimaging, only very limited public data exist (though see 6 for one example). Previously we shared the largest epilepsy neuroimaging data ever made publicly available -the Imaging Database for Epilepsy And Surgery (IDEAS, 7 ). That database contains T1weighted and FLAIR MRI acquisitions, additional to resection masks generated from post-operative imaging. Rich clinical/demographic metadata are also included, and IDEAS has already facilitated several further studies [8][9][10] .
Though valuable, the IDEAS data in its original release does not allow the inference of white matter structural networks or microstructural properties. This is an important limitation, since several studies demonstrate white matter alterations in epilepsy 11 , their importance for planning surgical treatment and mitigating post-operative sequelae [12][13][14][15] , and complementarity to neurophysiology when localizing epileptic activity 16 .
Diffusion weighted MRI (DWI) allows the inference of microstructural properties and white matter networks. Furthermore, DWI is recommended in epilepsy during pre-surgical evaluation to identify key tracts for surgical avoidance 17,18 . However, acquisition protocols vary widely, and DWI can be prone to technical artifacts which, if not addressed, can lead to major problems.
In this work, we build on our previous data release by sharing data from 216 people with drug resistant focal epilepsy and 98 healthy controls. We term this the IDEAS II data. All DWI scans were acquired using one of two acquisition protocols, with healthy controls acquired with each protocol. Fully preprocessed data are also available, as are structural connectivity matrices.
Data released in this paper are from a subset of the same subjects in our previous IDEAS release, the selection criteria for which have been described previously 7 . Included in this release are those subjects for whom DWI data was acquired pre-operatively and passed quality control.
For 83 (95) patients (controls) the DWI data and IDEAS T1w data were acquired during the same session, an example of which (‘sub-1’) is shown in figure 1. For the remainder, the DWI and T1w scans were acquired in different sessions. In total 123 (3) patients (controls) had DWI before the T1w scan, and 10 (0) with DWI after the T1w scan. When data were derived from multiple scan sessions, the interval was less than 24 months in 69% of individuals, and less than 36 months in 81% of individuals. In IDEASII, if scans were acquired in different sessions, these are denoted ‘ses-1’ and ‘ses-2’. Note that T1w and FLAIR scans shared in IDEAS II are identical to those already shared in IDEAS I.
Clinical and demographic information are included in our previous data release and are identical. Released metadata includes age, sex, histopathology, surgical outcomes, and seizure history. Almost all control scans (98 of 100) released previously are shared with DWI in the current release, and includes age and sex information.
Scans were acquired using one of two acquisition protocols described previously 19 . Scan details are included in the data release as javascript object notation (JSON) files, and are summarised below.
Protocol 1 data (N=167) were collected between 2014 and 2019 using a 3T GE MR750 scanner, equipped with a body coil for transmission and a 32-channel phased array coil for reception. Standard imaging gradients with a maximum strength of 50 mTm-1 and slew rate 200 Tm-1s-1 were fitted. Diffusionweighted MRI data were acquired using a single-shot echo planar imaging (EPI) sequence with echo time = 74.1ms and repetition time 7600ms. Sets of 70 contiguous 2-mm-thick axial slices were obtained covering the whole brain. A total of 115 volumes were acquired with 11, 8, 32, and 64 gradient directions at b-values of 0, 300, 700, and 2500/mm2, respectively (𝛿 = 21.5ms, 𝛥 = 35.9ms). The field of view was 25.6 × 25.6cm, and the acquisition matrix size was 128 × 128, giving a reconstructed voxel size of 2 × 2 × 2mm.
Protocol 2 data (N=147) were collected between 2009 and 2013 using a 3T GE Signa HDx scanner equipped with an eight-channel phased array coil. Diffusion MRI was collected using a cardiac-triggered single-shot EPI acquisition [TE = 73ms, TR = heart-rate dependent, b-value of 1200s/mm2 (𝛿 = 21ms, 𝛥 = 29ms, using maximum gradient strength of 40 mT m-1), 52 directions with 6 b0. Overall, 60 axial slices were collected, each 2.4-mm thick with a 96 × 96 matrix, zero-filled to 128 × 128 giving, 1.875 × 1.875mm in-plane resolution].
DWI data are frequently prone to artifacts such as EPI-induced distortions (see green arrow indications in figure 1), signal drift, and eddy-current induced distortions. To mitigate these, we release a fully preprocessed version of the data in addition to the raw (unprocessed
This content is AI-processed based on open access ArXiv data.