Quantitative CT radiomics-based models for prediction of haematoma expansion and poor functional outcome in primary intracerebral haemorrhage

Abstract

Objectives To test radiomics-based features extracted from noncontrast CT of patients with spontaneous intracerebral haemorrhage for prediction of haematoma expansion and poor functional outcome and compare them with radiological signs and clinical factors. Materials and methods Seven hundred fifty-four radiomics-based features were extracted from 1732 scans derived from the TICH-2 multicentre clinical trial. Features were harmonised and a correlation-based feature selection was applied. Different elastic-net parameterisations were tested to assess the predictive performance of the selected radiomics-based features using grid optimisation. For comparison, the same procedure was run using radiological signs and clinical factors separately. Models trained with radiomics-based features combined with radiological signs or clinical factors were tested. Predictive performance was evaluated using the area under the receiver operating characteristic curve (AUC) score. Results The optimal radiomics-based model showed an AUC of 0.693 for haematoma expansion and an AUC of 0.783 for poor functional outcome. Models with radiological signs alone yielded substantial reductions in sensitivity. Combining radiomics-based features and radiological signs did not provide any improvement over radiomics-based features alone. Models with clinical factors had similar performance compared to using radiomics-based features, albeit with low sensitivity for haematoma expansion. Performance of radiomics-based features was boosted by incorporating clinical factors, with time from onset to scan and age being the most important contributors for haematoma expansion and poor functional outcome prediction, respectively. Conclusion Radiomics-based features perform better than radiological signs and similarly to clinical factors on the prediction of haematoma expansion and poor functional outcome. Moreover, combining radiomics-based features with clinical factors improves their performance. Key Points • Linear models based on CT radiomics-based features perform better than radiological signs on the prediction of haematoma expansion and poor functional outcome in the context of intracerebral haemorrhage. • Linear models based on CT radiomics-based features perform similarly to clinical factors known to be good predictors. However, combining these clinical factors with radiomics-based features increases their predictive performance.

Location-Specific Radiomics Score: Novel Imaging Marker for Predicting Poor Outcome of Deep and Lobar Spontaneous Intracerebral Hemorrhage

“…On the other hand, the location-specific Rad-score that took into account the potential interaction between Rad-score and haematoma location can help predict poor outcomes at 6 months. Most recently, several studies suggested that NCCT radiomics features of haematomas could improve the predictive performance of prognostic models ( Pszczolkowski et al, 2021 ; Song et al, 2021b ). Nevertheless, the differences in radiomics features by haematoma location were not considered in these models, which may be valuable for proof-of-concept studies ( Anderson et al, 2010 ; Fu et al, 2014 ; Qureshi and Qureshi, 2018 ).…”

Development and Validation of a Clinical-Based Signature to Predict the 90-Day Functional Outcome for Spontaneous Intracerebral Hemorrhage

“…Detailed classification of hypodensity signs may improve accuracy; however, this had limited clinical applicability among doctors with insufficient clinical experience. Several recent studies ( Nawabi et al, 2021 ; Pszczolkowski et al, 2021 ; Song et al, 2021 ) have shown that radiomics, which applies machine learning (ML) algorithm, may be a more efficient and objective method for evaluating outcomes in patients with ICH. The ML method for evaluating the prognosis of patients with ICH remains in its infancy and has unstable models.…”

“…A review ( Morotti et al, 2019 ) was conducted regarding the detection, interpretation, reporting, and status of these markers, which may extensively improve their clinical application. With the advent of artificial intelligence (AI), radiomic methods have been extensively applied to predict the prognosis and HE of patients with ICH ( Shen et al, 2018 ; Xie et al, 2020 ; Xu et al, 2020 ; Pszczolkowski et al, 2021 ; Song et al, 2021 ). However, this method remains in the research phase; moreover, its predictive performance should be further verified.…”

Prediction of Intraparenchymal Hemorrhage Progression and Neurologic Outcome in Traumatic Brain Injury Patients Using Radiomics Score and Clinical Parameters

“…Therefore, the application of radiomics can extract more objective features that are not retrievable based on vision and manual measurements. It has consistently been shown to be effective for the prediction of hematoma expansion and poor neurologic outcome in spontaneous IPH patients [ 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 ]. As for traumatic IPH, we herein demonstrated its usefulness to further expand the limited evidence currently available [ 31 , 32 ].…”

“…Machine learning models generated from selected radiomics features have been shown to correlate well with clinically important diagnostic or prognostic outcomes [ 20 ]. Focusing on spontaneous intracerebral hemorrhages, extensive studies have been performed, which demonstrated the good predictive capability of models derived from radiomics features for hematoma expansion [ 21 , 22 , 23 , 24 , 25 ] and poor neurological outcome [ 24 , 26 , 27 , 28 , 29 , 30 ]. Furthermore, increased c-statistics were observed after the radiomics features were combined with clinical and radiological parameters constantly.…”