Abstract
BACKGROUND: Transcranial Doppler (TCD) has been used to estimate ICP noninvasively (nICP); however, its accuracy varies depending on different types of intracranial hypertension. Given the high specificity of TCD to detect cerebrovascular events, this study aimed to compare four TCD-based nICP methods during plateau waves of ICP. METHODS: A total of 36 plateau waves were identified in 27 patients (traumatic brain injury) with TCD, ICP, and ABP simultaneous recordings. The nICP methods were based on: (1) interaction between flow velocity (FV) and ABP using a "black-box" mathematical model (nICP_BB); (2) diastolic FV (nICP_FV (d) ); (3) critical closing pressure (nICP_CrCP), and (4) pulsatility index (nICP_PI). Analyses focused on relative changes in time domain between ICP and noninvasive estimators during plateau waves and the magnitude of changes (∆ between baseline and plateau) in real ICP and its estimators. A ROC analysis for an ICP threshold of 35 mmHg was performed. RESULTS: In time domain, nICP_PI, nICP_BB, and nICP_CrCP presented similar correlations: 0.80 ± 0.24, 0.78 ± 0.15, and 0.78 ± 0.30, respectively. nICP_FV (d) presented a weaker correlation (R = 0.62 ± 0.46). Correlations between ∆ICP and ∆nICP were better represented by nICP_CrCP and BB, R = 0.48, 0.44 (p < 0.05), respectively. nICP_FV (d) and PI presented nonsignificant ∆ correlations. ROC analysis showed moderate to good areas under the curve for all methods: nICP_BB, 0.82; nICP_FV (d) , 0.77; nICP_CrCP, 0.79; and nICP_PI, 0.81. CONCLUSIONS: Changes of ICP in time domain during plateau waves were replicated by nICP methods with strong correlations. In addition, the methods presented high performance for detection of intracranial hypertension. However, absolute accuracy for noninvasive ICP assessment using TCD is still low and requires further improvement.