Abstract：
For many years, noninvasive methods to measure intracranial pressure (ICP) have been unsuccessful. However, such methods are crucial for the assessment of patients with nonpenetrating traumatic brain injuries (TBIs) who are unconscious. In this study, we explored the use of transcranial transmission ultrasound (TTUS) to gather experimental data through brain pulsatility, assessing its effectiveness in detecting high ICP using machine learning analysis. We included patients with severe TBI under invasive ICP monitoring in our intensive care unit. During periods of both normal and elevated ICP, we simultaneously recorded ICP, arterial blood pressure, heart rate, and TTUS measurements. Our classification model was based on data from 9 patients, encompassing 387 instances of elevated ICP (>15 mmHg) and 345 instances of normal ICP (<10 mmHg), and validated through a leave-one-subject-out method. The study, conducted from October 2021 to October 2022, involved 25 patients with an average age of 61.6 ± 17.6 years, producing 279 datasets with an average ICP of 11.3 mmHg (1st quartile 6.1 mmHg; 3rd quartile 14.8 mmHg). The automated TTUS analysis effectively identified ICP values over 15 mmHg with 100% sensitivity and 47% specificity. It achieved a 100% negative predictive value and a 14% positive predictive value. This suggests that TTUS can accurately rule out high ICP above 15 mmHg in TBI patients, indicating patients who may need immediate imaging or intervention. These promising results, if confirmed and expanded in larger studies, could lead to the first reliable, noninvasive screening tool for detecting elevated ICP.
摘要：
多年来,非侵入性的方法来测量颅内压（ICP）没有成功。然而,这些方法对于评估无意识的非穿透性创伤性脑损伤（TBI）患者至关重要。在这项研究中,我们探索了使用经颅传导超声（TTUS）通过脑搏动来收集实验数据，使用机器学习分析评估其检测高ICP的有效性。我们纳入了重症监护病房中接受侵入性ICP监测的严重TBI患者。在正常和升高的ICP期间，我们同时记录ICP,动脉血压,心率,和TTUS测量。我们的分类模型基于9名患者的数据，包括387例ICP升高（>15mmHg）和345例ICP正常（<10mmHg），并通过留一法验证。这项研究,从2021年10月到2022年10月进行，涉及25名患者，平均年龄为61.6±17.6岁，产生279个数据集，平均ICP为11.3mmHg（第1四分位数6·1mmHg；第3四分位数14·8mmHg）。自动化TTUS分析以100%的灵敏度和47%的特异性有效地识别超过15mmHg的ICP值。它实现了100%的阴性预测值和14%的阳性预测值。这表明TTUS可以准确排除TBI患者高于15mmHg的高ICP，提示可能需要立即成像或干预的患者。这些有希望的结果，如果在更大的研究中得到证实和扩展，可能导致第一个可靠的，用于检测ICP升高的非侵入性筛查工具。
