Noninvasive intracranial pressure assessment based on a data-mining approach using a nonlinear mapping function

Artificial Intelligence Technologies in Neurosurgery: a Systematic Literature Review Using Topic Modeling. Part II: Research Objectives and Perspectives

The current increase in the number of publications on the use of artificial intelligence (AI) technologies in neurosurgery indicates a new trend in clinical neuroscience.

The aim of the study was to conduct a systematic literature review to highlight the main directions and trends in the use of AI in neurosurgery.

Artifact rejection and missing data imputation in cerebral blood flow velocity signals via trace norm minimization

OBJECTIVE

Many physiological signals are degraded by significant corruptions that limit their usefulness. One example is cerebral blood flow velocity (CBFV) signals, measured by transcranial Doppler, which are susceptible to large errors from patient motion. In this paper, we propose a method to remove artifacts and impute sections of missing data in these signals.

APPROACH

The method exploits the low-order dynamical relationship between CBFV, arterial blood pressure and, where available, intracranial pressure. It enhances the measured signals by fitting them to a low-order dynamical model, using convex regularization terms that improve robustness to large deviations and missing data. The method is based on a convex optimization formulation and utilizes recent work in trace norm approximation and subspace system identification.

MAIN RESULTS

Simulations demonstrate that the method successfully removes real CBFV artifacts and can impute missing data with reasonable accuracy. Performance was improved when intracranial pressure data was available.

CONCLUSION

The methods presented can be used by researchers to remove artifacts and estimate missing sections in CBFV signals. The general approach may be applied to other biomedical signal processing settings.

SIGNIFICANCE

This low-order dynamical approach has ongoing applications in noninvasive intracranial pressure estimation.

Intracranial Pressure Monitoring Signals After Traumatic Brain Injury: A Narrative Overview and Conceptual Data Science Framework

Continuous intracranial pressure (ICP) monitoring is a cornerstone of neurocritical care after severe brain injuries such as traumatic brain injury and acts as a biomarker of secondary brain injury. With the rapid development of artificial intelligent (AI) approaches to data analysis, the acquisition, storage, real-time analysis, and interpretation of physiological signal data can bring insights to the field of neurocritical care bioinformatics. We review the existing literature on the quantification and analysis of the ICP waveform and present an integrated framework to incorporate signal processing tools, advanced statistical methods, and machine learning techniques in order to comprehensively understand the ICP signal and its clinical importance. Our goals were to identify the strengths and pitfalls of existing methods for data cleaning, information extraction, and application. In particular, we describe the use of ICP signal analytics to detect intracranial hypertension and to predict both short-term intracranial hypertension and long-term clinical outcome. We provide a well-organized roadmap for future researchers based on existing literature and a computational approach to clinically-relevant biomedical signal data.

Intracranial Pressure and Intracranial Elastance Monitoring in Neurocritical Care

Patients with acute brain injuries tend to be physiologically unstable and at risk of rapid and potentially life-threatening decompensation due to shifts in intracranial compartment volumes and consequent intracranial hypertension. Invasive intracranial pressure (ICP) monitoring therefore remains a cornerstone of modern neurocritical care, despite the attendant risks of infection and damage to brain tissue arising from the surgical placement of a catheter or pressure transducer into the cerebrospinal fluid or brain tissue compartments. In addition to ICP monitoring, tracking of the intracranial capacity to buffer shifts in compartment volumes would help in the assessment of patient state, inform clinical decision making, and guide therapeutic interventions. We review the anatomy, physiology, and current technology relevant to clinical management of patients with acute brain injury and outline unmet clinical needs to advance patient monitoring in neurocritical care.

Review: pathophysiology of intracranial hypertension and noninvasive intracranial pressure monitoring

Measurement of intracranial pressure (ICP) is crucial in the management of many neurological conditions. However, due to the invasiveness, high cost, and required expertise of available ICP monitoring techniques, many patients who could benefit from ICP monitoring do not receive it. As a result, there has been a substantial effort to explore and develop novel noninvasive ICP monitoring techniques to improve the overall clinical care of patients who may be suffering from ICP disorders. This review attempts to summarize the general pathophysiology of ICP, discuss the importance and current state of ICP monitoring, and describe the many methods that have been proposed for noninvasive ICP monitoring. These noninvasive methods can be broken down into four major categories: fluid dynamic, otic, ophthalmic, and electrophysiologic. Each category is discussed in detail along with its associated techniques and their advantages, disadvantages, and reported accuracy. A particular emphasis in this review will be dedicated to methods based on the use of transcranial Doppler ultrasound. At present, it appears that the available noninvasive methods are either not sufficiently accurate, reliable, or robust enough for widespread clinical adoption or require additional independent validation. However, several methods appear promising and through additional study and clinical validation, could eventually make their way into clinical practice.

Algorithm for Reliable Detection of Pulse Onsets in Cerebral Blood Flow Velocity Signals

Transcranial Doppler (TCD) ultrasound has been demonstrated to be a valuable tool for assessing cerebral hemodynamics via measurement of cerebral blood flow velocity (CBFV), with a number of established clinical indications. However, CBFV waveform analysis depends on reliable pulse onset detection, an inherently difficult task for CBFV signals acquired via TCD. We study the application of a new algorithm for CBFV pulse segmentation, which locates pulse onsets in a sequential manner using a moving difference filter and adaptive thresholding. The test data set used in this study consists of 92,012 annotated CBFV pulses, whose quality is representative of real world data. On this test set, the algorithm achieves a true positive rate of 99.998% (2 false negatives), positive predictive value of 99.998% (2 false positives), and mean temporal offset error of 6.10 ± 4.75 ms. We do note that in this context, the way in which true positives, false positives, and false negatives are defined caries some nuance, so care should be taken when drawing comparisons to other algorithms. Additionally, we find that 97.8% and 99.5% of onsets are detected within 10 and 30 ms, respectively, of the true onsets. The algorithm's performance in spite of the large degree of variation in signal quality and waveform morphology present in the test data suggests that it may serve as a valuable tool for the accurate and reliable identification of CBFV pulse onsets in neurocritical care settings.

Real-Time Noninvasive Intracranial State Estimation Using Unscented Kalman Filter

Intracranial pressure (ICP) monitoring is desirable as a first-line measure to assist decision-making in cases of increased ICP. Clinically, non-invasive ICP monitoring is also required to avoid infection and hemorrhage in patients. The relationships among the arterial blood pressure (ABP), ICP, cerebral blood flow, and its velocity ( [Formula: see text]) measured by transcranial Doppler ultrasound measurement have been reported. However, real-time non-invasive ICP estimation using these modalities is less well documented. This paper presents a novel algorithm for real-time and non-invasive ICP monitoring with [Formula: see text] and ABP, called direct-current (DC)-ICP. The technique was compared with invasive ICP for 10 acute-brain-injury patients admitted to Cheju Halla Hospital and Gangnam Severance Hospital from July 2017 to June 2018. The inter-subject correlation coefficient between true and estimate was 0.75 and the AUCs of the ROCs for prediction of increased ICP for the DC-ICP methods were 0.83. Thus, [Formula: see text] monitoring can facilitate reliable real-time ICP tracking with our novel DC-ICP algorithm, which can provide valuable information under clinical conditions.

Identification of Pulse Onset on Cerebral Blood Flow Velocity Waveforms: A Comparative Study

The low cost, simple, noninvasive, and continuous measurement of cerebral blood flow velocity (CBFV) by transcranial Doppler is becoming a common clinical tool for the assessment of cerebral hemodynamics. CBFV monitoring can also help with noninvasive estimation of intracranial pressure and evaluation of mild traumatic brain injury. Reliable CBFV waveform analysis depends heavily on its accurate beat-to-beat delineation. However, CBFV is inherently contaminated with various types of noise/artifacts and has a wide range of possible pathological waveform morphologies. Thus, pulse onset detection is in general a challenging task for CBFV signal. In this paper, we conducted a comprehensive comparative analysis of three popular pulse onset detection methods using a large annotated dataset of 92,794 CBFV pulses—collected from 108 subarachnoid hemorrhage patients admitted to UCLA Medical Center. We compared these methods not only in terms of their accuracy and computational complexity, but also for their sensitivity to the selection of their parameters' values. The results of this comprehensive study revealed that using optimal values of the parameters obtained from sensitivity analysis, one method can achieve the highest accuracy for CBFV pulse onset detection with true positive rate (TPR) of 97.06% and positive predictivity value (PPV) of 96.48%, when error threshold is set to just less than 10 ms. We conclude that the high accuracy and low computational complexity of this method (average running time of 4ms/pulse) makes it a reliable algorithm for CBFV pulse onset detection.

Data-Augmented Modeling of Intracranial Pressure

Precise management of patients with cerebral diseases often requires intracranial pressure (ICP) monitoring, which is highly invasive and requires a specialized ICU setting. The ability to noninvasively estimate ICP is highly compelling as an alternative to, or screening for, invasive ICP measurement. Most existing approaches for noninvasive ICP estimation aim to build a regression function that maps noninvasive measurements to an ICP estimate using statistical learning techniques. These data-based approaches have met limited success, likely because the amount of training data needed is onerous for this complex applications. In this work, we discuss an alternative strategy that aims to better utilize noninvasive measurement data by leveraging mechanistic understanding of physiology. Specifically, we developed a Bayesian framework that combines a multiscale model of intracranial physiology with noninvasive measurements of cerebral blood flow using transcranial Doppler. Virtual experiments with synthetic data are conducted to verify and analyze the proposed framework. A preliminary clinical application study on two patients is also performed in which we demonstrate the ability of this method to improve ICP prediction.

Non-Invasive Electromagnetic Skin Patch Sensor to Measure Intracranial Fluid-Volume Shifts

Elevated intracranial fluid volume can drive intracranial pressure increases, which can potentially result in numerous neurological complications or death. This study’s focus was to develop a passive skin patch sensor for the head that would non-invasively measure cranial fluid volume shifts. The sensor consists of a single baseline component configured into a rectangular planar spiral with a self-resonant frequency response when impinged upon by external radio frequency sweeps. Fluid volume changes (10 mL increments) were detected through cranial bone using the sensor on a dry human skull model. Preliminary human tests utilized two sensors to determine feasibility of detecting fluid volume shifts in the complex environment of the human body. The correlation between fluid volume changes and shifts in the first resonance frequency using the dry human skull was classified as a second order polynomial with R² = 0.97. During preliminary and secondary human tests, a ≈24 MHz and an average of ≈45.07 MHz shifts in the principal resonant frequency were measured respectively, corresponding to the induced cephalad bio-fluid shifts. This electromagnetic resonant sensor may provide a non-invasive method to monitor shifts in fluid volume and assist with medical scenarios including stroke, cerebral hemorrhage, concussion, or monitoring intracranial pressure.

Non-invasive Monitoring of Intracranial Pressure Using Transcranial Doppler Ultrasonography: Is It Possible?

Although intracranial pressure (ICP) is essential to guide management of patients suffering from acute brain diseases, this signal is often neglected outside the neurocritical care environment. This is mainly attributed to the intrinsic risks of the available invasive techniques, which have prevented ICP monitoring in many conditions affecting the intracranial homeostasis, from mild traumatic brain injury to liver encephalopathy. In such scenario, methods for non-invasive monitoring of ICP (nICP) could improve clinical management of these conditions. A review of the literature was performed on PUBMED using the search keywords 'Transcranial Doppler non-invasive intracranial pressure.' Transcranial Doppler (TCD) is a technique primarily aimed at assessing the cerebrovascular dynamics through the cerebral blood flow velocity (FV). Its applicability for nICP assessment emerged from observation that some TCD-derived parameters change during increase of ICP, such as the shape of FV pulse waveform or pulsatility index. Methods were grouped as: based on TCD pulsatility index; aimed at non-invasive estimation of cerebral perfusion pressure and model-based methods. Published studies present with different accuracies, with prediction abilities (AUCs) for detection of ICP ≥20 mmHg ranging from 0.62 to 0.92. This discrepancy could result from inconsistent assessment measures and application in different conditions, from traumatic brain injury to hydrocephalus and stroke. Most of the reports stress a potential advantage of TCD as it provides the possibility to monitor changes of ICP in time. Overall accuracy for TCD-based methods ranges around ±12 mmHg, with a great potential of tracing dynamical changes of ICP in time, particularly those of vasogenic nature.

A systematic study of linear dynamic modeling of intracranial pressure dynamics

Our group has proposed a generic time series data mining framework and demonstrated its potential as a noninvasive intracranial pressure (ICP) assessment approach. The linear dynamic model (LDM) was used in our previous work without rigorous justification. In the current study, we performed a systematic study of the practical performance of the LDM for ICP dynamics by investigating three important aspects to consider in using the LDM to model ICP dynamics. Those three aspects include the fitness of the LDM to data, the generalizability of the models, and the choice of input signals to the models. Our study results show that the fitness of the LDM to data is excellent and the LDM for ICP dynamics is well generalizable, which is of particular interest to adopting our time series data mining framework for noninvasive ICP assessment.