Detection of small bleeds in the brain with electrical impedance tomography

Abstract

In this paper, we describe and assess feasibility of instrumentation and algorithms for detecting bleeding due to hemorrhagic strokes and traumatic brain injury using electrical impedance tomography, a novel biomedical diagnostic modality in which the body is probed noninvasively with generally imperceptible alternating currents applied in patterns to a set of electrodes placed in contact with the skin. We focus on the GENESIS instrument developed by GE Global Research and on the achievability of our goal to detect a bleed in the center of the head with a volume of several ml. Our main topic is compensation for the large changes in voltages that tend to occur when the electrodes are in contact with biological media, specifically either human subjects or with vegetable matter proxies which seem to exhibit the same 'drift' phenomenon. We show that these changes in voltages can be modeled by assuming that each electrode is attached to the body via a discrete complex impedance whose value is time-varying and describe how this discrete component value can be estimated and largely compensated-for. We compare this discrete model with changes in contact impedances estimated using the complete electrode model showing that the two models are roughly comparable in their ability to explain the data from a single human subject experiment with electrodes attached to the head. In a simulation study, we demonstrate that it is possible to detect a small bleed in the center of the head even in the case of large changes in electrode impedances, which can be treated as nuisance parameters.

Advances in electrical impedance tomography-based brain imaging

“…Xiuzhen Dong's group [80,89] achieved real-time detection and continuous monitoring of subsequent pathological states by establishing subarachnoid hemorrhage and ICH models in piglets. Other groups have conducted studies on the detection of small bleeding in the brain, intraventricular hemorrhage, and the use of support vector machine classifiers for detecting cerebral hemorrhage [90][91][92].…”

A Review of Algorithms and Hardware Implementations in Electrical Impedance Tomography (Invited)

“…In [56], the supervised descent method (SDM) is used to learn the sensitive matrix J as shown in Eq. (5). Compared with the traditional iterative method in model-based approach, SDM lessens the time and memory cost notably, and the priori knowledge is assimilated in a more pliable manner.…”

“…Compared with other tomography technologies [1], such as computed tomography (CT), gamma ray tomography, X-ray tomography, ultrasonic imaging, and magnetic resonance imaging (MRI), EIT is a real-time, noninvasive, low-cost, and portable technology which can reflect the internal structures and functions of tissues and organs [2]. Furthermore, it has shown the ability of long-term monitoring of diseases in clinical practice, such as lung function imaging [3,4], cerebral hemorrhage [5], early detection of breast cancer [6,7], and gastric emptying detection [8]. Especially during the novel coronavirus (COVID-19), Tomasino and others used EIT technology to evaluate the distribution of ventilation in the different pulmonary regions before, during, and after pronation in COVID-19 respiratory failure in real time [9].…”

Tomography for electrical impedance

“…It should be highlighted that EIT, is a non-invasive technique that does not generate radiation [56], [57], providing greater patient safety, although not yet widely studied as a possible method for monitoring joint injuries. This technique allows for the evaluation, diagnosis and detection of diseases such as: pneumothorax [58], brain disorders [59], [60], lung diseases in which both the effects of positive end-expiratory pressure are determined [61], [62], flow of gas in regions of the lungs [63], [64] and regional changes in the ventilation distribution [65], [66] based on the distribution of conductivity; making this technique a viable alternative in the study of joints [49], [51].…”