Prospective Ocular Imaging for Intracranial Pressure Evaluation

Overview

This is a prospectively recruiting, database development study collecting images and videos of the spontaneous venous pulsation at the back of people's eyes - this is a pulse one can see on examination of the back of the eye, originating from the blood vessels around the nerve that connects the eye to the brain (the optic nerve), and is present in most people who have normal pressure around the brain. However, in people with raised pressure in the brain, this pulse disappears as the pressure rises. Many things can cause the pressure around the brain to increase, including tumours, swellings and trauma. The investigators want to test if high-quality images and videos of this pulse, taken using both hand-held and larger, fixed-platform machines, can be used to train a software tool to automatically detect this pulse. The investigators want to collect these images and videos in 2 groups of patients: those with no known or suspected brain pressure problems, and those who are suspected to have raised pressure and/or are due to undergo measurement of the pressure around the brain, called lumbar punctures or intracranial pressure bolt monitoring. These tests to check the pressure around the brain are invasive - they involve inserting needles in the back or directly into the brain to measure the pressure, and carry risks. The value of these two groups of people will be to help train the software to reasonably say whether a pulse is present or absent and, hopefully, estimate what the pressure around the brain may be without the need for an invasive test.

Raised ICP occurs in many serious neurological conditions including idiopathic intracranial hypertension (IIH), subarachnoid haemorrhage, haemorrhagic cerebrovascular accidents (CVA), meningitis, intracranial tumours and traumatic brain injury (TBI), and if undiagnosed or untreated can lead to blindness, brain injury and death. The need to accurately determine ICP is a frequent clinical dilemma in the fields of ophthalmology, neurology, neurosurgery, and emergency and critical care medicine, and traditionally has relied on a combination of clinical evaluation and measurement with lumbar punctures (LPs), or the surgical insertion of transcranial pressure transducers (ICP bolts), with or without additional computerised tomography (CT) and magnetic resonance imaging (MRI) scans which are often required in cases of suspected acutely elevated ICP. These invasive tests are the only direct methods of measuring the ICP, and carry risks of pain, post-procedural headaches and cerebrospinal fluid (CSF) leaking, intracranial haemorrhage, infection, neuronal and cortical injury, and rarely death. There are associated health service and patient costs, as these tests are cumulatively expensive and generally require inpatient admission and monitoring (with the exception of relatively few units well-versed with and resourced for day-case LPs). Given these risks, the evaluation of pragmatic, non-invasive methods of ICP measurement that are acceptable to patients have attracted increasing interest, which have included ultrasonic optic nerve sheath diameter measurement, transcranial doppler, pupillometry and tympanic membrane displacement, among others. The presence or absence of SVP on funduscopic examination has been a long used by clinicians to evaluate the likelihood of raised ICP. First described by Coccius, early mechanistic in-human studies found that SVPs disappear once the ICP rises above approximately 20 +/- 2.5 cmH2O. On the basis of clinician determination alone (by fundus examination), SVPs have been estimated to be present in 70-80% of all eyes and 80-90% of all individuals with presumably normal ICP, with a sensitivity of 0.89 and positive predictive value of 0.88 for excluding raised ICP. SVPs classically involve only short segments of retinal vein, usually on or in close proximity to the rim of the optic disc, and are theorised to arise due to differences in the rise and fall of intraocular pressure (IOP) and cerebrospinal fluid (CSF) pulse pressure (PP) with the cardiac cycle. Under normal physiological states, the mean retinal venous pressure (RVP) is consistently higher than mean IOP, which maintains continuous ocular venous outflow. As retinal veins are thin-walled and lack rigidity, fluctuations in surrounding pressure are directly transmitted to the vessels. The IOP rises and falls by approximately 1.5 mmHg in systole and diastole respectively, and this pressure variance is transmitted directly into the retinal veins. The retinal veins converge into the central retinal vein (CRV) which passes through the lamina cribrosa with the optic nerve, then exits the nerve sheath 10 mm behind the globe, entering the subarachnoid space where it becomes subject to changes in CSF PP. Unlike IOP, CSF PP only rises and falls by 0.5 mmHg during systole and diastole, thus in systole the transmitted rise in IOP causes the RVP to 1 mmHg higher than CSF PP relative to its normal pressure differential, and the opposite happens in diastole. This has the effect of causing the portion of the CRV which is most subject to this pressure differential (usually proximal to the optic disc) to collapse during systole as blood outflow increases, and to expand in diastole as blood outflow decreases, according to Poiseuille's law which states that flow rate is directly proportional to the pressure gradient. This effect diminishes the further the vein is away from the disc where it is dampened by surrounding structures, explaining why SVPs are generally most visible in short segments of veins, in close proximity to the disc. In situations of raised ICP, dampening and loss of SVPs are theorised to occur due to a rise in CSF PP that increases in a linear fashion with ICP. Eventually as ICP approaches or exceeds 20 cmH2O, the CSF PP matches that of the IOP pulse pressure, causing SVPs to disappear. The visibility of the SVP can also change independent of ICP, for example in situations of retinal vein occlusions and arterial occlusions which can abnormally alter RVP and directly injure retinal vessels leading to diminution of a visible SVP. SVPs have also been reported as appearing less frequently in patients with open-angle glaucoma (OAG) and normal-tension glaucoma (NTG) with decreasing SVP visibility associated with increasing functional field loss, though the exact mechanism is unclear. Orbit-related pathology such as thyroid eye disease (TED) has also been reported to cause absence of SVPs, presumably due to the external compressive effect on the optic nerve causing raised retrolaminar venous pressure. Furthermore, as noted above, 10% (or more) of individuals in adult populations with no known issues related to intracranial pressure may have an absent SVP determined through clinician examination alone, although it is well reported that high inter-observer variability and the general inclusion of glaucomatous patients in earlier cohorts may have confounded this finding. A final consideration is the effect of patient posture on SVP visibility. Studies that describe SVPs typically have patients sat erect and upright, postured on a slit-lamp or fundus image capture device, however it has been demonstrated that changing posture from an upright to lateral decubitus position (which raises the ICP by \~8-10 cmH2O) can reduce SVP amplitude by much as \~18% in healthy participants with visible SVPs. Studies utilising video capture of optic disc pulsations, for example using OCT devices, have shown that high fidelity optic disc recordings can increase the detection rate of SVPs up to 99% in healthy adult patients with otherwise normal optic discs, increasing the viable utility of SVPs as a biomarker to estimate ICP. Additional advantages of using video capture devices include the capacity to record captured video allowing for detection, verification, and quantification of SVPs, the option to use different light wavelengths to improve vessel contrast and visibility (such as infra-red and red-free imaging), and improved resolution thanks to sophisticated optics and post-processing software. Studies have shown that OCT videos compared to standard fundus slit-lamp examination improve the visibility of SVPs from 48.6% to 86.7% of patients, particularly for patients with more subtle SVPs, with a high inter-rater reliability of 0.82 (Cohen a). The same studies have also reported significantly higher mean 24-hour ICP and 24-hour pulse amplitude in patients with no SVP compared to patients with SVP present, and a significant association between higher ICP and reduced SVP grade. The main drawbacks of wider use of OCT devices to non-invasively assess ICP include the limited portability of the devices, the training required to acquire and interpret high-quality images, and difficulties capturing images in patients who cannot sit upright. To overcome the limitations of video capture on fixed-position OCT devices, hand-held fundus video capture devices are being increasingly developed and evaluated, either as purpose-built standalone video capture platforms (such as the EpiCam or Zeiss Visuscout) or in the form of smartphone/device attachments to ophthalmoscopes (such as the PanOptic Plus Ophthalmoscope, PPO). The above study, which reported reduction in SVP amplitude of 18% comparing upright to lateral decubitus posturing, measured SVP amplitude as a function of difference in venous pixel density between maximal and minimal vessel diameter on videos recorded by a tablet-mounted ophthalmoscope capturing at 30 frames per second (fps). This allows for SVP detection in individuals who would not otherwise be able to posture at a slit-lamp for clinician assessment. The main drawback to this and other studies of hand-held devices has been the lack of data and quantification of SVP metrics in individuals with raised ICP, absence of comparison to more established optic disc video capture devices such as OCT, and the need for expert clinician analysis or post-processing. Additionally, advances in video capture technology have led to recent significant improvements in video quality with current generation smartphones able to capture 4K resolution videos, which provide high-fidelity videos of the fundus. The capabilities of modern imaging hardware have allowed better visualisation of retina and optic disc than ever before, even using hand-held devices. Therefore, there exists scope for an automated tool that can be used to detect, localise and quantify SVPs. This may, in turn, feed into the development of automated tools to estimate contemporaneous ICP using these imaging data, which would have applications in the diagnosis and monitoring of disorders associated with raised ICP. This may potentially spare patients from currently required invasive tests of ICP measurement including LPs or intracranial bolts, in hospital and community settings.