Skip to main content

Research Repository

Advanced Search

Semantic Segmentation of Spontaneous Intracerebral Hemorrhage, Intraventricular Hemorrhage, and Associated Edema on CT Images Using Deep Learning

Kok, Yong En; Pszczolkowski, Stefan; Law, Zhe Kang; Ali, Azlinawati; Krishnan, Kailash; Bath, Philip; Sprigg, Nikola; Dineen, Robert A.; French, Andrew P.

Authors

Yong En Kok

Zhe Kang Law

Azlinawati Ali

Kailash Krishnan

PHILIP BATH philip.bath@nottingham.ac.uk
Stroke Association Professor of Stroke Medicine

NIKOLA SPRIGG nikola.sprigg@nottingham.ac.uk
Professor of Stroke Medicine

Andrew P. French



Abstract

This study evaluated deep learning algorithms for semantic segmentation and quantification of intracerebral hemorrhage (ICH), perihematomal edema (PHE), and intraventricular hemorrhage (IVH) on noncontrast CT scans of patients with spontaneous ICH. Models were assessed on 1732 annotated baseline noncontrast CT scans obtained from the Tranexamic Acid for Hyperacute Primary Intracerebral Haemorrhage (ie, TICH-2) international multicenter trial (ISRCTN93732214), and different loss functions using a three-dimensional no-new-U-Net (nnU-Net) were examined to address class imbalance (30% of participants with IVH in dataset). On the test cohort (n = 174, 10% of dataset), the top-performing models achieved median Dice similarity coefficients of 0.92 (IQR, 0.89–0.94), 0.66 (0.58–0.71), and 1.00 (0.87–1.00), respectively, for ICH, PHE, and IVH segmentation. U-Net–based networks showed comparable, satisfactory performances on ICH and PHE segmentations (P . .05), but all nnU-Net variants achieved higher accuracy than the Brain Lesion Analysis and Segmentation Tool for CT (BLAST-CT) and DeepLabv31 for all labels (P, .05). The Focal model showed improved performance in IVH segmentation compared with the Tversky, two-dimensional nnU-Net, U-Net, BLAST-CT, and DeepLabv31 models (P, .05). Focal achieved concordance values of 0.98, 0.88, and 0.99 for ICH, PHE, and ICH volumes, respectively. The mean volumetric differences between the ground truth and prediction were 0.32 mL (95% CI: −8.35, 9.00), 1.14 mL (−9.53, 11.8), and 0.06 mL (−1.71, 1.84), respectively. In conclusion, U-Net–based networks provide accurate segmentation on CT images of spontaneous ICH, and Focal loss can address class imbalance. International Clinical Trials Registry Platform (ICTRP) no. ISRCTN93732214.

Citation

Kok, Y. E., Pszczolkowski, S., Law, Z. K., Ali, A., Krishnan, K., Bath, P., …French, A. P. (2022). Semantic Segmentation of Spontaneous Intracerebral Hemorrhage, Intraventricular Hemorrhage, and Associated Edema on CT Images Using Deep Learning. Radiology: Artificial Intelligence, 4(6), https://doi.org/10.1148/ryai.220096

Journal Article Type Article
Acceptance Date Sep 12, 2022
Online Publication Date Sep 28, 2022
Publication Date 2022-11
Deposit Date Sep 14, 2022
Publicly Available Date Mar 29, 2023
Journal Radiology: Artificial Intelligence
Print ISSN 2638-6100
Electronic ISSN 2638-6100
Publisher Radiological Society of North America
Peer Reviewed Peer Reviewed
Volume 4
Issue 6
DOI https://doi.org/10.1148/ryai.220096
Keywords Artificial Intelligence; Radiology, Nuclear Medicine and imaging; Radiological and Ultrasound Technology
Public URL https://nottingham-repository.worktribe.com/output/11195449
Publisher URL https://pubs.rsna.org/doi/10.1148/ryai.220096