Accelerating Prediction of Malignant Cerebral Edema After Ischemic Stroke with Automated Image Analysis and Explainable Neural Networks.
Brain computed tomography scan
Brain edema
Cerebral infarction
Deep learning
Early diagnosis
Journal
Neurocritical care
ISSN: 1556-0961
Titre abrégé: Neurocrit Care
Pays: United States
ID NLM: 101156086
Informations de publication
Date de publication:
04 2022
04 2022
Historique:
received:
30
03
2021
accepted:
02
08
2021
pubmed:
22
8
2021
medline:
5
4
2022
entrez:
21
8
2021
Statut:
ppublish
Résumé
Malignant cerebral edema is a devastating complication of stroke, resulting in deterioration and death if hemicraniectomy is not performed prior to herniation. Current approaches for predicting this relatively rare complication often require advanced imaging and still suffer from suboptimal performance. We performed a pilot study to evaluate whether neural networks incorporating data extracted from routine computed tomography (CT) imaging could enhance prediction of edema in a large diverse stroke cohort. An automated imaging pipeline retrospectively extracted volumetric data, including cerebrospinal fluid (CSF) volumes and the hemispheric CSF volume ratio, from baseline and 24 h CT scans performed in participants of an international stroke cohort study. Fully connected and long short-term memory (LSTM) neural networks were trained using serial clinical and imaging data to predict those who would require hemicraniectomy or die with midline shift. The performance of these models was tested, in comparison with regression models and the Enhanced Detection of Edema in Malignant Anterior Circulation Stroke (EDEMA) score, using cross-validation to construct precision-recall curves. Twenty of 598 patients developed malignant edema (12 required surgery, 8 died). The regression model provided 95% recall but only 32% precision (area under the precision-recall curve [AUPRC] 0.74), similar to the EDEMA score (precision 28%, AUPRC 0.66). The fully connected network did not perform better (precision 33%, AUPRC 0.71), but the LSTM model provided 100% recall and 87% precision (AUPRC 0.97) in the overall cohort and the subgroup with a National Institutes of Health Stroke Scale (NIHSS) score ≥ 8 (p = 0.0001 vs. regression and fully connected models). Features providing the most predictive importance were the hemispheric CSF ratio and NIHSS score measured at 24 h. An LSTM neural network incorporating volumetric data extracted from routine CT scans identified all cases of malignant cerebral edema by 24 h after stroke, with significantly fewer false positives than a fully connected neural network, regression model, and the validated EDEMA score. This preliminary work requires prospective validation but provides proof of principle that a deep learning framework could assist in selecting patients for surgery prior to deterioration.
Sections du résumé
BACKGROUND
Malignant cerebral edema is a devastating complication of stroke, resulting in deterioration and death if hemicraniectomy is not performed prior to herniation. Current approaches for predicting this relatively rare complication often require advanced imaging and still suffer from suboptimal performance. We performed a pilot study to evaluate whether neural networks incorporating data extracted from routine computed tomography (CT) imaging could enhance prediction of edema in a large diverse stroke cohort.
METHODS
An automated imaging pipeline retrospectively extracted volumetric data, including cerebrospinal fluid (CSF) volumes and the hemispheric CSF volume ratio, from baseline and 24 h CT scans performed in participants of an international stroke cohort study. Fully connected and long short-term memory (LSTM) neural networks were trained using serial clinical and imaging data to predict those who would require hemicraniectomy or die with midline shift. The performance of these models was tested, in comparison with regression models and the Enhanced Detection of Edema in Malignant Anterior Circulation Stroke (EDEMA) score, using cross-validation to construct precision-recall curves.
RESULTS
Twenty of 598 patients developed malignant edema (12 required surgery, 8 died). The regression model provided 95% recall but only 32% precision (area under the precision-recall curve [AUPRC] 0.74), similar to the EDEMA score (precision 28%, AUPRC 0.66). The fully connected network did not perform better (precision 33%, AUPRC 0.71), but the LSTM model provided 100% recall and 87% precision (AUPRC 0.97) in the overall cohort and the subgroup with a National Institutes of Health Stroke Scale (NIHSS) score ≥ 8 (p = 0.0001 vs. regression and fully connected models). Features providing the most predictive importance were the hemispheric CSF ratio and NIHSS score measured at 24 h.
CONCLUSIONS
An LSTM neural network incorporating volumetric data extracted from routine CT scans identified all cases of malignant cerebral edema by 24 h after stroke, with significantly fewer false positives than a fully connected neural network, regression model, and the validated EDEMA score. This preliminary work requires prospective validation but provides proof of principle that a deep learning framework could assist in selecting patients for surgery prior to deterioration.
Identifiants
pubmed: 34417703
doi: 10.1007/s12028-021-01325-x
pii: 10.1007/s12028-021-01325-x
pmc: PMC8858326
mid: NIHMS1737973
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Langues
eng
Sous-ensembles de citation
IM
Pagination
471-482Subventions
Organisme : NINDS NIH HHS
ID : U24 NS107230
Pays : United States
Organisme : NINDS NIH HHS
ID : P30 NS098577
Pays : United States
Organisme : NINDS NIH HHS
ID : K23 NS099487
Pays : United States
Organisme : NINDS NIH HHS
ID : K23 NS099440
Pays : United States
Organisme : NINDS NIH HHS
ID : R01 NS085419
Pays : United States
Informations de copyright
© 2021. Springer Science+Business Media, LLC, part of Springer Nature and Neurocritical Care Society.
Références
Hacke W, Schwab S, Horn M, et al. “Malignant” middle cerebral artery territory infarction: clinical course and prognostic signs. Arch Neurol. 1996;53:309–15.
doi: 10.1001/archneur.1996.00550040037012
Vahedi K, Hofmeijer J, Juettler E, et al. Early decompressive surgery in malignant infarction of the middle cerebral artery: a pooled analysis of three randomised controlled trials. Lancet Neurol. 2007;6:215–22.
doi: 10.1016/S1474-4422(07)70036-4
Liebeskind DS, Juttler E, Shapovalov Y, et al. Cerebral Edema associated with large hemispheric infarction. Stroke. 2019;50(9):2619–25.
doi: 10.1161/STROKEAHA.118.024766
Wu S, Yuan R, Wang Y, et al. Early prediction of malignant brain edema after ischemic stroke. Stroke. 2018;49(12):2918–27.
doi: 10.1161/STROKEAHA.118.022001
MacCallum C, Churilov L, Mitchell P, Dowling R, Yan B. Low Alberta Stroke Program Early CT score (ASPECTS) associated with malignant middle cerebral artery infarction. Cerebrovasc Dis. 2014;38(1):39–45.
doi: 10.1159/000363619
Thomalla G, Hartmann F, Juettler E, et al. Prediction of malignant middle cerebral artery infarction by magnetic resonance imaging within 6 hours of symptom onset: a prospective multicenter observational study. Ann Neurol. 2010;68:435–45.
doi: 10.1002/ana.22125
Bosche B, Hamann GF, Dohmen C, Graf R. There is more to it than: the greater the infarction volume, the more probable is a malignant MCA infarction. Stroke. 2006;37(3):762–3.
doi: 10.1161/01.STR.0000204054.30458.e8
Dohmen C, Bosche B, Graf R, et al. Prediction of malignant course in MCA infarction by PET and microdialysis. Stroke J Cerebral Circ. 2003;34:2152–8.
doi: 10.1161/01.STR.0000083624.74929.32
Burghaus L, Liu WC, Dohmen C, et al. Prognostic value of electroencephalography and evoked potentials in the early course of malignant middle cerebral artery infarction. Neurol Sci. 2013;34(5):671–8.
doi: 10.1007/s10072-012-1102-1
Minnerup J, Wersching H, Ringelstein EB, et al. Prediction of malignant middle cerebral artery infarction using computed tomography-based intracranial volume reserve measurements. Stroke. 2011;42:3403–9.
doi: 10.1161/STROKEAHA.111.619734
Kauw F, Bennink E, de Jong H, et al. Intracranial cerebrospinal fluid volume as a predictor of malignant middle cerebral artery infarction. Stroke. 2019. https://doi.org/10.1161/STROKEAHA.119.026853 .
doi: 10.1161/STROKEAHA.119.026853
pubmed: 31451100
pmcid: 6553515
Ong CJ, Gluckstein J, Laurido-Soto O, et al. Enhanced detection of edema in malignant anterior circulation stroke (EDEMA) score: a risk prediction tool. Stroke. 2017;48(7):1969–72.
doi: 10.1161/STROKEAHA.117.016733
Dhar R, Yuan K, Kulik T, et al. CSF volumetric analysis for quantification of cerebral edema after hemispheric infarction. Neurocrit Care. 2016;24(3):420–7.
doi: 10.1007/s12028-015-0204-z
Chen Y, Dhar R, Heitsch L, et al. Automated quantification of cerebral edema following hemispheric infarction: application of a machine-learning algorithm to evaluate CSF shifts on serial head CTs. Neuroimage Clin. 2016;12:673–80.
doi: 10.1016/j.nicl.2016.09.018
Dhar R, Chen Y, Hamzehloo A, et al. Reduction in cerebrospinal fluid volume as an early quantitative biomarker of cerebral edema after ischemic stroke. Stroke. 2020;51(2):462–7.
doi: 10.1161/STROKEAHA.119.027895
Foroushani HM, Hamzehloo A, Kumar A, et al. Quantitative serial CT imaging-derived features improve prediction of malignant cerebral edema after ischemic stroke. Neurocrit Care. 2020. https://doi.org/10.1007/s12028-020-01056-5 .
doi: 10.1007/s12028-020-01056-5
pubmed: 32729090
pmcid: 7738418
Dhar R, Hamzehloo A, Kumar A, et al. Hemispheric CSF volume ratio quantifies progression and severity of cerebral edema after acute hemispheric stroke. J Cereb Blood Flow Metab. 2021. https://doi.org/10.1177/0271678X211018210 .
doi: 10.1177/0271678X211018210
pubmed: 34013805
Dhar R, Chen Y, An H, Lee JM. Application of machine learning to automated analysis of cerebral edema in large cohorts of ischemic stroke patients. Front Neurol. 2018;9:687.
doi: 10.3389/fneur.2018.00687
Rorden C, Bonilha L, Fridriksson J, Bender B, Karnath HO. Age-specific CT and MRI templates for spatial normalization. Neuroimage. 2012;61(4):957–65.
doi: 10.1016/j.neuroimage.2012.03.020
Miao J, Song X, Sun W, et al. Predictors of malignant cerebral edema in cerebral artery infarction: a meta-analysis. J Neurol Sci. 2020;409:116607.
doi: 10.1016/j.jns.2019.116607
Van Houdt G, Mosquera C, Napoles G. A review on the long short-term memory model. Artif Intell Rev. 2020;53(8):5929–55.
doi: 10.1007/s10462-020-09838-1
Ge W, Huh JW, Park YR, et al. An interpretable ICU mortality prediction model based on logistic regression and recurrent neural networks with LSTM units. AMIA Ann Symp Proc. 2018;2018:460–9.
Shung D, Huang J, Castro E, et al. Neural network predicts need for red blood cell transfusion for patients with acute gastrointestinal bleeding admitted to the intensive care unit. Sci Rep. 2021;11(1):8827.
doi: 10.1038/s41598-021-88226-3
Sun Y, Kaur R, Gupta S, et al. Development and validation of high definition phenotype-based mortality prediction in critical care units. JAMIA Open. 2021;4(1):ooab004.
doi: 10.1093/jamiaopen/ooab004
Pediaditis M, Ciubotaru AN, Brunschwiler T, et al. Machine learning techniques for personalized detection of epileptic events in clinical video recordings. AMIA Ann Symp Proc. 2020;2020:1003–11.
Kwak GH, Ling L, Hui P. Predicting the need for vasopressors in the intensive care unit using an attention based deep learning model. Shock. 2021;56(1):73–79.
doi: 10.1097/SHK.0000000000001692
Ge Y, Wang Q, Wang L, et al. Predicting post-stroke pneumonia using deep neural network approaches. Int J Med Inf. 2019;132:103986.
doi: 10.1016/j.ijmedinf.2019.103986
Saito T, Rehmsmeier M. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS One. 2015;10(3):e0118432.
doi: 10.1371/journal.pone.0118432
Lauritsen SM, Kristensen M, Olsen MV, et al. Explainable artificial intelligence model to predict acute critical illness from electronic health records. Nat Commun. 2020;11(1):3852.
doi: 10.1038/s41467-020-17431-x
Cheng Y, Wu S, Wang Y, et al. External validation and modification of the EDEMA score for predicting malignant brain EDEMA after acute ischemic stroke. Neurocrit Care. 2020;32(1):104–12.
doi: 10.1007/s12028-019-00844-y
Broocks G, Kemmling A, Meyer L, et al. Computed tomography angiography collateral profile is directly linked to early EDEMA progression rate in acute ischemic stroke. Stroke. 2019;50(12):3424–30.
doi: 10.1161/STROKEAHA.119.027062
Berger L, Hakim AM. The association of hyperglycemia with cerebral edema in stroke. Stroke. 1986;17(5):865–71.
doi: 10.1161/01.STR.17.5.865
Kirsch E, Szejko N, Falcone GJ. Genetic underpinnings of cerebral edema in acute brain injury: an opportunity for pathway discovery. Neurosci Lett. 2020;730:135046.
doi: 10.1016/j.neulet.2020.135046
Wijdicks EFM, Sheth KN, Carter BS, et al. Recommendations for the management of cerebral and cerebellar infarction with swelling: a statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke J Cerebral Circ. 2014;45:1222–38.
doi: 10.1161/01.str.0000441965.15164.d6
Dhar R. Automated quantitative assessment of cerebral edema after ischemic stroke using CSF volumetrics. Neurosci Lett. 2020;724:134879.
doi: 10.1016/j.neulet.2020.134879
Thorsen-Mayer H-C, Nielson AB, Nielsen AP, et al. Dynamic and explainable machine learning prediction of mortality in patients in the intensive care unit: a retrospective study of high-frequency data in electronic patient records. Lancet Digital Health. 2020. https://doi.org/10.1016/S2589-7500(20)30018-2 .
doi: 10.1016/S2589-7500(20)30018-2
Thoren M, Azevedo E, Dawson J, et al. Predictors for cerebral edema in acute ischemic stroke treated with intravenous thrombolysis. Stroke. 2017;48(9):2464–71.
doi: 10.1161/STROKEAHA.117.018223
Finlayson SG, Subbaswamy A, Singh K, et al. The clinician and dataset shift in artificial intelligence. N Engl J Med. 2021;385(3):283–6.
doi: 10.1056/NEJMc2104626
Lundberg SM, Nair B, Vavilala MS, et al. Explainable machine-learning predictions for the prevention of hypoxaemia during surgery. Nat Biomed Eng. 2018;2(10):749–60.
doi: 10.1038/s41551-018-0304-0
Bentley P, Ganesalingam J, Carlton Jones AL, et al. Prediction of stroke thrombolysis outcome using CT brain machine learning. Neuroimage Clin. 2014;4:635–40.
doi: 10.1016/j.nicl.2014.02.003