Journal article
Characterizing clot composition and stroke etiology through hounsfield units
Neuroradiology
09/23/2026
DOI: 10.1007/s00234-026-04193-4
PMID: 42776213
Appears in UI Libraries Support Open Access
Abstract
We aimed to analyze clot composition using Hounsfield Units (HU) in non-enhanced computed tomography (NECT).
Thrombi retrieved from mechanical thrombectomy (MT) were imaged on micro-computed tomography (micro-CT) and histologically analyzed. Micro-CT slides were matched to histological sections. Spearman's rank correlation was performed to correlate micro-CT HU with red blood cells (RBCs) and fibrin composition. A sensitivity analysis was conducted between micro-CT HU and histological slides with > 70% of these components. The same clots were then localized in the pre-MT NECT, and NECT HU values were obtained. Micro-CT HU and NECT HU were correlated for histologically informed interpretation. Receiver operating characteristic analysis was performed to retrieve NECT HU thresholds. A dataset of patients who presented with large vessel occlusion of cardioembolic, large artery atherosclerosis (LAA), and cryptogenic strokes were then assessed through NECT HU.
Ten thrombi were analyzed. Micro-CT HU were strongly correlated with histological sections with > 70% of RBCs (Rho 0.675) but not to sections with > 70% of fibrin (Rho 0.325). Micro-CT HU and NECT HU of clots that had > 70% of RBCs were strongly correlated (Rho 0.774). NECT HU of 44 had a sensitivity of 67% and specificity of 86% to determine clots with > 70% of RBCs. A total of 240 NECT images were analyzed including 80 cardioembolic, 80 LAA and 80 cryptogenic strokes. NECT HU assessment showed that 75% of cardioembolic, 40% of LAA and 55% of cryptogenic clots are composed of RBCs.
NECT HU is a promising tool to assess clot RBCs composition and infer stroke etiology.
Details
- Title: Subtitle
- Characterizing clot composition and stroke etiology through hounsfield units
- Creators
- Andres Gudino - University of IowaCarlos Dier - University of IowaMartin Cabarique - University of IowaAlexander Van Dam - University of IowaAlina Rivadeneira-Limongi - University of Iowa, NeurologyEmily Baniewicz - University of IowaMario S Hinojosa-Figueroa - University of Iowa, NeurologyArshaq Saleem - University of IowaDivanshu Sharma - University of IowaNavami Shenoy - University of IowaElena Sagues - University of IowaArathi Ashok - University of IowaSusan A Walsh - University of IowaMalik Ghannam - University of IowaOsorio Lopes Abath Neto - University of IowaEdgar A Samaniego - University of Iowa
- Resource Type
- Journal article
- Publication Details
- Neuroradiology
- DOI
- 10.1007/s00234-026-04193-4
- PMID
- 42776213
- NLM abbreviation
- Neuroradiology
- ISSN
- 1432-1920
- eISSN
- 1432-1920
- Publisher
- Springer Nature
- Grant note
- NIH grant: 520 1S10ODO18503 01, CSS-CNV-23-001
Imaging acquisition on the micro-CT scanner was supported by the NIH grant 520 1S10ODO18503 01. This study was supported by the grant CSS-CNV-23-001 awarded by Johnson and Johnson - Neuro.
- Language
- English
- Electronic publication date
- 09/23/2026
- Academic Unit
- Neurology; Radiology; Pathology; Iowa Neuroscience Institute; Neurosurgery
- Record Identifier
- 9985236322002771
Metrics
1 Record Views