Ji Soo-Yeon, Najarian Kayvan, Huynh Toan, Jeong Dong Hyun
Department of Computer Science, Bowie State University, 14000 Jericho Park Road, Bowie, MD, 20715, USA.
Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, 2800 Plymouth Road, Ann Arbor, MI, 48109, USA.
Methods Mol Biol. 2017;1598:405-419. doi: 10.1007/978-1-4939-6952-4_21.
In Traumatic Brain Injury (TBI), elevated Intracranial Pressure (ICP) causes severe brain damages due to hemorrhage and swelling. Monitoring ICP plays an important role in the treatment of TBI patients because ICP is considered a strong predictor of neurological outcome and a potentially amenable method to treat patients. However, it is difficult to predict and measure accurate ICP due to the complex nature of patients' clinical conditions. ICP monitoring for severe TBI patient is a challenging problem for clinicians because traditionally known ICP monitoring is an invasive procedure by placing a device inside the brain to measure pressure. Therefore, ICP monitoring might have a high infection risk and cause medical complications. In here, an ICP monitoring using texture features is proposed to overcome this limitation. The combination of image processing methods and a decision tree algorithm is utilized to estimate ICP of TBI patients noninvasively. In addition, a visual analytics tool is used to conduct an interactive visual factor analysis and outlier detection.
在创伤性脑损伤（TBI）中，颅内压（ICP）升高会因出血和肿胀导致严重的脑损伤。监测颅内压在TBI患者的治疗中起着重要作用，因为颅内压被认为是神经功能预后的有力预测指标，也是一种潜在可行的治疗患者的方法。然而，由于患者临床状况的复杂性，很难预测和准确测量颅内压。对重症TBI患者进行颅内压监测对临床医生来说是一个具有挑战性的问题，因为传统的颅内压监测是通过在脑内放置设备来测量压力的侵入性操作。因此，颅内压监测可能有很高的感染风险并引发医疗并发症。在此，提出了一种利用纹理特征进行颅内压监测的方法来克服这一局限性。利用图像处理方法和决策树算法的组合来无创估计TBI患者的颅内压。此外，还使用了一种视觉分析工具进行交互式视觉因素分析和异常值检测。
