Normalized CT perfusion maps from 62 SAH patients

Kobbai, Alexandre1 ; Mesnildrey, Quentin 1 ; Szabo, Vivien1 ; Chalard, Kévin 1

Published Aug 13, 2025 on Dryad. https://doi.org/10.5061/dryad.0zpc86784

Data files

Aug 13, 2025 version files 7.85 GB

Abstract

Aneurysmal subarachnoid hemorrhage (SAH) is a severe condition that triggers numerous metabolic disruptions and complications. While the exact mechanisms implied in this cascade of events are still being investigated, monitoring cerebral perfusion seems critical to understand and prevent the occurrence of secondary deficits such as delayed cerebral ischemia (DCI).

The present dataset was built within the framework of a retrospective observational study in SAH patients.

Sixty-two patients were included. Clinical information at admission, aneurysm details, neurological events, and rescue treatments were collected. Of note, within this study population, 33% of patients were classified as WFNS III-V. Cerebral vasospasm (CVS) occurrence was 68%, and that of DCI was 15%.

Overall, 873 CT perfusion parametric maps (TMAX, MTT, CBF) were collected and preprocessed. In particular, all data were normalized to the MNI (Montreal Neurological Institute) standardized space that allows precise within and between-subject analyses. Normalized cerebral perfusion maps can also be analyzed in specific anatomical or arterial regions of interest using the relevant segmentation templates.

In this project, we developed an image processing pipeline that allowed quantitative analysis of CTP parameters' dynamics.

https://doi.org/10.5061/dryad.0zpc86784

CONTRIBUTORS:

-Alexandre Kobbai

-Quentin Mesnildrey (corresponding author: quentin.mesnildrey@chu-montpellier.fr)

-Vivien Szabo

-Kévin Chalard

RESEARCH SCOPE:

Evolution of cerebral perfusion in patients diagnosed with aneurysmal subarachnoid hemorrhage.

Description of the data and file structure

This dataset includes:

873 preprocessed CT perfusion maps (TMAX, CBF, MTT) acquired in 62 patients diagnosed with aSAH.

Data were initially acquired using GE Medical Systems device (24-slice scanner, GE Healthcare, Milwaukee, Wisconsin) with a slice thickness of 5mm.

Parametric maps were computed using Perfusion_4D software (General Electric, revo_ct_22bc.50). The basilar artery was manually selected as the arterial input function and the venous outflow was automatically determined by the software.

DICOM raw data were converted to nifti format and registered to the MNI standardized space (size 181217181).

patients_table.xlsx

Excel file containing two sheets ('Patients' and 'Variables description').

An .xls file (table patients_table.xls) containing clinical information for each patient (eg: Subject ID, Age, Sex, BMI, Neurological scores, Aneurysm details, clinical events and dates relative to the first symptoms, treatments and procedure used, etc). Additional details are provided in the 'Variables description' sheet.

'Variables description' sheet describes the nature of all variables collected in 'Patients' sheet. 'Patients' sheet contains one row per patient (ie 62 rows), with column 1 listing all patients identifier (ID).

normalizedCTP.zip

Contains one folder per patient and named after their unic identifier (ID) (ie 62 folders in total). Each folder contains the MNI-registered CT perfusion parametric maps in NIFTI (.nii) format.

Filenames are formatted as follows: <subjectID>_exam<exam number>CT<metric>_w.nii

<subjectID> refers to subjects' unic identifiers (ID in patients_table.xlsx, starting at S003)

<exam number> refers to the different CTP exams acquired in a given patient in chronological order.

<metric> can be: 'tmax', 'cbf' or 'mtt'.

the '_w' suffix indicates that the corresponding data was normalized ('warped') to the standardized MNI space.

Each file is a 3D volume of size: dim1xdim2xdim3 = 181x217x181 voxels. Voxel size is 1x1x1mm3.

dim1 = left to right side of the patient's head.
dim2 = posterior to anterior side of the patient's head.
dim3 = bottom to top of the patient's head.

CTP_exams_dates.json

JSON structure file containing one field per patient (ie, 62 fields). Each field contains an array of float numbers representing the delay, in days, between each CTP exam and the first SAH symptoms (Day 0). The length of each array, therefore, matches the number of exams for a given patient.

Human subjects data

The present dataset does not contain any patient-related information. Subject identifiers are arbitrary; name, sex, and date of exam were removed from the metadata. Exam dates provided in the CTP_exam_dates.json file represent the delay relative to the initial SAH symptoms. All data were acquired in accordance with French ethics and regulations regarding patients' information, consent, data curation, and publication.