BLAST-CT in the browser

The traumatic brain injury lesion model of the University of Cambridge and Imperial College London (Monteiro et al., Lancet Digit Health 2020): intraparenchymal and extra-axial haemorrhage, perilesional oedema and intraventricular haemorrhage on non-contrast head CT. Use it for SAH after a fall or for traumatic ICH. Everything runs on this computer.

Research and education use only

A research demonstration, not a medical device. This model is not validated or approved for clinical use here; do not use it for diagnosis, treatment or any other clinical decision. Segmentations can be wrong; always review them against the images.

Your images stay on this computer

Files are read and processed inside this browser tab on your GPU or CPU. No image data or results are uploaded. Only the page, the model files and the demo cases are downloaded, and the model is cached for later visits.

What you need

  • A recent desktop Chrome or Edge (Firefox and Safari may work on the CPU).
  • 8 GB RAM or more. WebGPU is fastest; on the CPU a scan takes one to several minutes.
  • A non-contrast axial head CT in Hounsfield units.
Compute
B

BLAST-CT Cambridge & Imperial · traumatic brain injury lesions

Four lesion classes from the CENTER-TBI work: intraparenchymal haemorrhage, extra-axial haemorrhage, perilesional oedema and intraventricular haemorrhage. Use it for SAH after a fall (traumatic SAH is part of the extra-axial class) or for traumatic ICH.

Menon, Newcombe, Glocker et al. · University of Cambridge NeuroICU · Imperial College London
Models
About 2 MB per model, cached.
Not the same task as SAHVAI and CLAIM: EAH combines subarachnoid, subdural and extradural blood, and the model was trained on traumatic brain injury, not aneurysmal SAH.

Drop a head CT here

A zip of a DICOM study, a DICOM folder, loose DICOM files, a .nii / .nii.gz or a .nrrd volume. The scan is resampled to 1 mm isotropic, as in the blast-ct tool.

reference: none · must share the scan's geometry; Dice is reported per class

Citation, licence and disclaimer

Model. Monteiro M, Newcombe VFJ, Mathieu F, Adatia K, Kamnitsas K, Ferrante E, Das T, Whitehouse D, Rueckert D, Menon DK, Glocker B. Multiclass semantic segmentation and quantification of traumatic brain injury lesions on head CT using deep learning: an algorithm development and multicentre validation study. Lancet Digit Health 2020;2(6):e314–e322. doi:10.1016/S2589-7500(20)30085-6

Code and weights. github.com/biomedia-mira/blast-ct (BioMedIA, Imperial College London, with the University of Cambridge Division of Anaesthesia), Apache License 2.0, Copyright 2020 Imperial College London. The 12 pre-trained DeepMedic models are converted to ONNX (fp16) without retraining. Architecture: Kamnitsas K, Ledig C, Newcombe VFJ, et al. Med Image Anal 2017;36:61–78. The software "is not intended for clinical use" (blast-ct README). Demo: scan_0 and its label map from the blast-ct repository.

How it runs here. As blast-ct: reorientation, 1 mm isotropic linear resampling (SimpleITK-style), window −15 to 100 HU scaled to −1…1, three-pathway DeepMedic over tiles, softmax averaged over the chosen models, argmax. Output tiles of 60³ give the same labels as the tool's 30³ tiles (checked: 0 voxels differ).

Disclaimer. For research and education only. Not a medical device; not cleared or approved by the FDA, CE or any regulator; not validated for clinical use; must not be used for diagnosis, triage, treatment or any other clinical decision. Outputs can be wrong and must be reviewed against the images by a qualified reader. Provided "as is" without warranty of any kind. Images are processed locally in this browser and are not uploaded.

Log