Table 1 Overview of related work on segmentation in cranial NCCT.
From: Intracerebral Haemorrhage Segmentation in Non-Contrast CT
Author | Year | Method | nD | Pathology | Reference standard | Observer variability | Test set (N) | ||
|---|---|---|---|---|---|---|---|---|---|
Patients | Scans | Slices | |||||||
Bardera et al.29 | 2009 | Semi-automated region growing and level set | — | ICH and edema | Slice-based manual† | Inter | 18 | 18 | — |
Liao et al.30 | 2009 | Multi-resolution binary level set | 2D | ICH, SDH and EDH | Manual† | No | 48 | 48 | 48 |
Prakash et al.31 | 2012 | Modified distance regularised level set evolution | 2D | ICH and IVH | Slice-based manual | No | — | 50 | — |
Prakash et al.32 | 2012 | Modified thresholding, clustering, and graph theory | 2D | ICH and IVH | Slice-based manual | No | — | 51 | — |
Bhadauria et al.33 | 2013 | Fuzzy C-mean clustering and active contour | 2D | ICH, SDH and EDH | Manual† | No | 20 | — | 100 |
Boers et al.34 | 2014 | Atlas registration and region growing | 2D | SAH | Manual | Inter | 30 | 30 | — |
Gillebert et al.38 | 2014 | Template registration and statistical analysis | 2D | Stroke lesions | Slice-based manual | No | 24 | 24 | — |
Scherer et al.39 | 2016 | Random forest classifier | — | ICH | Manual/Semi-automatic | Inter | 30 | 30 | — |
Shahangian et al.35 | 2016 | Modified distance regularised level set evolution | 2D | ICH, IVH, SDH and EDH | Manual† | No | — | — | 627 |
Gautam et al.36 | 2018 | Fuzzy clustering and distance regularized level set evolution | 2D | ICH | Manual† | No | — | — | 60 |
Chang et al.37 | 2018 | Convolutional neural network† | 2D | ICH, SDH, EDH and SAH | Semi-automatic† | No | — | 862 (82) | 23,668 (—) |
Muschelli et al.40 | 2017 | Random forest classifier | 3D† | ICH | Manual/Semi-automatic | No | 102 | 102 | — |
This work | 2019 | Convolutional neural network | 3D | ICH | Manual | Inter & Intra | 30/— | 30/50 | — |
