Early prediction of clinical outcomes in patients with aneurysmal subarachnoid hemorrhage using computed tomography texture analysis

Abstract

Radiological evaluation of subarachnoid hemorrhage (SAH) is often subject to interobserver variability. The aim of this study was to retrospectively detect computed tomography (CT) texture parameters in the early postictal state to predict cerebral vasospasm, delayed cerebral ischemia (DCI), and functional outcome in aneurysmal SAH using quantitative CT texture analysis (CTTA) via a commercially available software program and routine CT images. 40 patients with aneurysmal SAH surgically treated at the Keio University Hospital during a four-year period were analyzed. CT texture analyses were performed using a commercially available software program (Synapse Vincent). The following texture parameters of blood clots in the subarachnoid space and cerebral edema were assessed: mean CT value, entropy, skewness, and kurtosis. The mean CT value of blood clots in the subarachnoid space was significantly associated with cerebral vasospasm, DCI, and functional outcome. The mean CT value ≥ 49.64 Hounsfield units (HU) predicted cerebral vasospasm with a sensitivity and specificity of 85.7% and 61.5%, respectively (area under the curve [AUC] = 0.758). The mean CT value ≥ 49.95 HU predicted DCI with a sensitivity and specificity of 100% and 60.6%, respectively (AUC = 0.810). The mean CT value ≥ 53.00 HU predicted poor functional outcome with a sensitivity and specificity of 56.3% and 91.7%, respectively (AUC = 0.747). CTTA using a commercially available software program demonstrated that the mean CT value of clots in the subarachnoid space in the early postictal state could predict vasospasm, DCI, and clinical outcome with a high sensitivity and specificity.