CTseg: A Tool for Brain CT Segmentation, Spatial Normalisation, and Volumetrics

CTseg, introduced by Mikael Brudfors and published on arXiv, is a freely available software for brain CT segmentation, spatial normalization, and volumetrics. It builds on the Multi-Brain generative modeling framework to provide a CT-specific pipeline that produces tissue maps, deformation fields, and brain volume estimates in the same format as SPM's unified segmentation. This allows researchers to apply established MRI analysis workflows to CT scans without needing to modify existing pipelines. A key design feature is that CTseg works directly on routine hospital CT scans without any preprocessing or resampling, making it practical for clinical deployment. The software is already used in studies on stroke, dementia, and brain morphometry, but until now lacked systematic validation against an independent reference standard.

To validate CTseg, the authors used paired MR and CT head scans from the same individuals, evaluating four dimensions: segmentation accuracy against an MRI-derived silver standard, spatial normalization consistency via group-average sharpness and voxelwise coefficient of variation, brain volume agreement via intraclass correlation and Bland-Altman analysis, and downstream sex classification performance from normalized tissue maps. They compared CTseg against directly applying SPM's MRI-based unified segmentation to CT images. Results showed CTseg significantly outperformed the baseline in segmentation and normalization, demonstrated stronger total brain volume agreement, and achieved comparable total intracranial volume agreement. CTseg is available on GitHub with full experiment code for reproducibility, offering a powerful, validated tool for clinical neuroimaging research on CT data.