1
|
Oliveira G, Fonseca AC, Ferro J, Oliveira AL. Deep Learning-Based Extraction of Biomarkers for the Prediction of the Functional Outcome of Ischemic Stroke Patients. Diagnostics (Basel) 2023; 13:3604. [PMID: 38132189 PMCID: PMC10743068 DOI: 10.3390/diagnostics13243604] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/13/2023] [Revised: 11/26/2023] [Accepted: 12/01/2023] [Indexed: 12/23/2023] Open
Abstract
Accurately predicting functional outcomes in stroke patients remains challenging yet clinically relevant. While brain CTs provide prognostic information, their practical value for outcome prediction is unclear. We analyzed a multi-center cohort of 743 ischemic stroke patients (<72 h onset), including their admission brain NCCT and CTA scans as well as their clinical data. Our goal was to predict the patients' future functional outcome, measured by the 3-month post-stroke modified Rankin Scale (mRS), dichotomized into good (mRS ≤ 2) and poor (mRS > 2). To this end, we developed deep learning models to predict the outcome from CT data only, and models that incorporate other patient variables. Three deep learning architectures were tested in the image-only prediction, achieving 0.779 ± 0.005 AUC. In addition, we created a model fusing imaging and tabular data by feeding the output of a deep learning model trained to detect occlusions on CT angiograms into our prediction framework, which achieved an AUC of 0.806 ± 0.082. These findings highlight how further refinement of prognostic models incorporating both image biomarkers and clinical data could enable more accurate outcome prediction for ischemic stroke patients.
Collapse
Affiliation(s)
- Gonçalo Oliveira
- NeuralShift, 1000-138 Lisbon, Portugal
- INESC-ID, Instituto Superior Técnico, 1000-029 Lisbon, Portugal
| | - Ana Catarina Fonseca
- Faculdade de Medicina, Universidade de Lisboa, 1649-028 Lisbon, Portugal; (A.C.F.); (J.F.)
| | - José Ferro
- Faculdade de Medicina, Universidade de Lisboa, 1649-028 Lisbon, Portugal; (A.C.F.); (J.F.)
| | | |
Collapse
|
2
|
Karimian-Jazi K, Vollherbst DF, Schwarz D, Fischer M, Schregel K, Bauer G, Kocharyan A, Sturm V, Neuberger U, Jesser J, Herweh C, Ulfert C, Hilgenfeld T, Seker F, Preisner F, Schmitt N, Charlet T, Hamelmann S, Sahm F, Heiland S, Wick W, Ringleb PA, Schirmer L, Bendszus M, Möhlenbruch MA, Breckwoldt MO. MR microscopy to assess clot composition following mechanical thrombectomy predicts recanalization and clinical outcome. J Neurointerv Surg 2023:jnis-2023-020594. [PMID: 37527928 DOI: 10.1136/jnis-2023-020594] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/17/2023] [Accepted: 07/16/2023] [Indexed: 08/03/2023]
Abstract
BACKGROUND Mechanical thrombectomy (MT) is the standard of care for patients with a stroke and large vessel occlusion. Clot composition is not routinely assessed in clinical practice as no specific diagnostic value is attributed to it, and MT is performed in a standardized 'non-personalized' approach. Whether different clot compositions are associated with intrinsic likelihoods of recanalization success or treatment outcome is unknown. METHODS We performed a prospective, non-randomized, single-center study and analyzed the clot composition in 60 consecutive patients with ischemic stroke undergoing MT. Clots were assessed by ex vivo multiparametric MRI at 9.4 T (MR microscopy), cone beam CT, and histopathology. Clot imaging was correlated with preinterventional CT and clinical data. RESULTS MR microscopy showed red blood cell (RBC)-rich (21.7%), platelet-rich (white,38.3%) or mixed clots (40.0%) as distinct morphological entities, and MR microscopy had high accuracy of 95.4% to differentiate clots. Clot composition could be further stratified on preinterventional non-contrast head CT by quantification of the hyperdense artery sign. During MT, white clots required more passes to achieve final recanalization and were not amenable to contact aspiration compared with mixed and RBC-rich clots (maneuvers: 4.7 vs 3.1 and 1.2 passes, P<0.05 and P<0.001, respectively), whereas RBC-rich clots showed higher probability of first pass recanalization (76.9%) compared with white clots (17.4%). White clots were associated with poorer clinical outcome at discharge and 90 days after MT. CONCLUSION Our study introduces MR microscopy to show that the hyperdense artery sign or MR relaxometry could guide interventional strategy. This could enable a personalized treatment approach to improve outcome of patients undergoing MT.
Collapse
Affiliation(s)
| | - Dominik F Vollherbst
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Daniel Schwarz
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Manuel Fischer
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Katharina Schregel
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Gregor Bauer
- Neurology Clinic, University Hospital Heidelberg, Heidelberg, Germany
| | - Anna Kocharyan
- Department of Neurology, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
| | - Volker Sturm
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Ulf Neuberger
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Jessica Jesser
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Christian Herweh
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Christian Ulfert
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Tim Hilgenfeld
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Fatih Seker
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Fabian Preisner
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Niclas Schmitt
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Tobias Charlet
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Stefan Hamelmann
- Department of Neuropathology, University of Heidelberg, Heidelberg, Germany
- Clinical Cooperation Unit Neuropathology, German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany
| | - Felix Sahm
- Department of Neuropathology, University of Heidelberg, Heidelberg, Germany
- Clinical Cooperation Unit Neuropathology, German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany
| | - Sabine Heiland
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Wolfgang Wick
- Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany
- Clinical Cooperation Unit Neurooncology, German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany
| | - Peter A Ringleb
- Neurology Clinic, University Hospital Heidelberg, Heidelberg, Germany
| | - Lucas Schirmer
- Department of Neurology, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
| | - Martin Bendszus
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Markus A Möhlenbruch
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| | - Michael O Breckwoldt
- Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany
| |
Collapse
|