Admission computed tomography radiomic signatures outperform hematoma volume in predicting baseline clinical severity and functional outcome in the ATACH‐2 trial intracerebral hemorrhage population. (18th July 2021)

• Record Type: Journal Article
• Authors: Haider, Stefan P.; Qureshi, Adnan I.; Jain, Abhi; Tharmaseelan, Hishan; Berson, Elisa R.; Zeevi, Tal; Majidi, Shahram; Filippi, Christopher G.; Iseke, Simon; Gross, Moritz; Acosta, Julian N.; Malhotra, Ajay; Kim, Jennifer A.; Sansing, Lauren H.; Falcone, Guido J.; Sheth, Kevin N.; Payabvash, Seyedmehdi
• Is Part Of: European journal of neurology. Volume 28:Number 9(2021)
• Page Start: 2989
• Page End: 3000
• Publication Date: 2021-07-18
• Subjects: hematoma -- intracerebral hemorrhage -- outcome -- radiomics -- volume; Neurology -- Periodicals; Nervous system -- Diseases -- Periodicals; 616.8
• DOI: 10.1111/ene.15000
• Languages: English
• ISSNs: 1351-5101

Abstract

Abstract: Background and purpose: Radiomics provides a framework for automated extraction of high‐dimensional feature sets from medical images. We aimed to determine radiomics signature correlates of admission clinical severity and medium‐term outcome from intracerebral hemorrhage (ICH) lesions on baseline head computed tomography (CT). Methods: We used the ATACH‐2 (Antihypertensive Treatment of Acute Cerebral Hemorrhage II) trial dataset. Patients included in this analysis ( n = 895) were randomly allocated to discovery ( n = 448) and independent validation ( n = 447) cohorts. We extracted 1130 radiomics features from hematoma lesions on baseline noncontrast head CT scans and generated radiomics signatures associated with admission Glasgow Coma Scale (GCS), admission National Institutes of Health Stroke Scale (NIHSS), and 3‐month modified Rankin Scale (mRS) scores. Spearman's correlation between radiomics signatures and corresponding target variables was compared with hematoma volume. Results: In the discovery cohort, radiomics signatures, compared to ICH volume, had a significantly stronger association with admission GCS (0.47 vs. 0.44, p = 0.008), admission NIHSS (0.69 vs. 0.57, p < 0.001), and 3‐month mRS scores (0.44 vs. 0.32, p < 0.001). Similarly, in independent validation, radiomics signatures, compared to ICH volume, had a significantly stronger association with admission GCS (0.43 vs. 0.41, p = 0.02), NIHSS (0.64 vs. 0.56, p < 0.001), and 3‐month mRS scores (0.43 vs. 0.33, p < 0.001). In multiple regression analysis adjusted for known predictors of ICH outcome, the radiomics signature was an independent predictor of 3‐month mRS in both cohorts. Conclusions: Limited by the enrollment criteria of the ATACH‐2 trial, we showed that radiomics features quantifying hematoma texture, density, and shape on baseline CT can provide imaging correlates for clinical presentation and 3‐month outcome. These findings couldtrigger a paradigm shift where imaging biomarkers may improve current modelsfor prognostication, risk‐stratification, and treatment triage of ICH patients. Abstract : Radiomics provides a framework for automated extraction of high‐dimensional feature sets from medical images quantifying lesion density, shape, and texture. Using the ATACH‐2 (Antihypertensive Treatment of Acute Cerebral Hemorrhage II) trial dataset, we applied volumetric segmentation of intracerebral hemorrhage lesions on noncontrast baseline computed tomography, image pre‐processing, and extraction of 1130 radiomic features to derive a comprehensive quantitative representation of hematoma imaging characteristics. In a discovery cohort, radiomics signatures for prediction of admission clinical severity and 3‐month outcome were devised by linearly combining sets of robust radiomics features exhibiting strong association with target variables while minimizing feature multicollinearity.