Risk Assessment for Neurobehavioral Toxicity

Background: Brain activity has been investigated by several methods with different principles, notably optical ones. Each method may offer information on distinct physiological or pathological aspects of brain function. The ideal instrument to measure brain activity should include complementary techniques and integrate the resultant information. As a "low cost" approach towards this objective, we combined the well-grounded electroencephalography technique with the newer near infrared spectroscopy methods to investigate human visual function. Methods: The article describes an embedded instrumentation combining a continuous-wave nearinfrared spectroscopy system and an electroencephalography system to simultaneously monitor functional hemodynamics and electrical activity. Near infrared spectroscopy (NIRS) signal depends on the light absorption spectra of haemoglobin and measures the blood volume and blood oxygenation regulation supporting the neural activity. The NIRS and visual evoked potential (VEP) are concurrently acquired during steady state visual stimulation, at 8 Hz, with a b/w "windmill" pattern, in nine human subjects. The pattern contrast is varied (1%, 10%, 100%) according to a stimulation protocol. Results: In this study, we present the measuring system; the results consist in concurrent recordings of hemodynamic changes and evoked potential responses emerging from different contrast levels of a patterned stimulus. The concentration of [HbO2] increases and [HHb] decreases after the onset of the stimulus. Their variation shows a clear relationship with the contrast value: large contrast produce huge difference in concentration, while low contrast provokes small concentration difference. This behaviour is similar to the already known relationship between VEP response amplitude and contrast. Conclusion: The simultaneous recording and analysis of NIRS and VEP signals in humans during visual stimulation with a b/w pattern at variable contrast, demonstrates a strong linear correlation between hemodynamic changes and evoked potential amplitude. Furthermore both responses present a logarithmic profile with stimulus contrast. Published: 4 July 2007 BioMedical Engineering OnLine 2007, 6:28 doi:10.1186/1475-925X-6-28 Received: 5 April 2007 Accepted: 4 July 2007 This article is available from: http://www.biomedical-engineering-online.com/content/6/1/28 © 2007 Rovati et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Page 1 of 15 (page number not for citation purposes) BioMedical Engineering OnLine 2007, 6:28 http://www.biomedical-engineering-online.com/content/6/1/28 Background Several methods to investigate brain activity have been proposed to date [1]. The most common techniques can be classified in terms of the task measured: neural activity or hemodynamics. Magnetoencephalography and electroencephalography measure neural activity directly via magnetic and electrical fields exhibiting excellent temporal resolution but poor spatial resolution [2,3]. In contrast to these methods, emission tomography, functional magnetic resonance and near-infrared spectroscopy measure hemodynamic changes associated with neural activation [4,5]. In particular, emission tomography measures cerebral metabolic rate of oxygen and cerebral blood flow, whereas functional magnetic resonance measures blood flow and blood oxygenation level dependent (BOLD) changes. Both these methods offer excellent spatial resolution but limited temporal performance. On the other hand, near-infrared spectroscopy overcomes the temporal resolution limitation of these techniques offering nonetheless intermediate spatial resolution performance. An important feature of all these techniques is their contrast-to-noise ratio, i.e. task-induced change relative to background noise. Each method has peculiarities that emphasize certain mechanisms involved in stimulus perception or interpretation, e.g. electrical/magnetic mechanisms, mechanical/thermal mechanisms or hemodynamic mechanisms. Many variables come into play in the gamut of brain activities, and thus the integration of different techniques may represent an appropriate strategy. Motion artefacts must be addressed for all these techniques, even if some methods, such as electroencephalography, are relatively less sensitive to motion as compared to others, e.g. the combination of electroencephalography and near infrared spectroscopy may provide a more reliable interpretation of brain activity. Over the last decades, great improvements have been achieved in the investigation of the human brain by these techniques [1-5]. Nevertheless, it would be advantageous to simultaneously observe both neural and hemodynamic activity. The design of such instrumentation might also encourage more cross talk among scientists studying the intricate complexities of this process. In this framework, we present a system combining a continuous-wave near infrared spectrometer (NIMO, NIROX srl, Italy) and an electroencephalography system. A single personal computer is used to record and process signals and to control the visual stimulation during the experiments. This design allows comparison between neural electrical activity (evoked potential signals) and hemodynamic response (near infrared spectroscopy) triggered by the same stimulation. After a discussion on the basic theory, measuring principle and system configuration, we present an application of the instrument through experimental results obtained from stimulation of the human visual system with different contrast levels. Methods A. Measuring principle Near infrared spectroscopy and electroencephalography have a complementary role in extracting spatial and temporal information on direct (electrical) or indirect (metabolic) neural activity. Near infrared spectroscopy, which measures the light absorption spectra of hemoglobin, directly measures changes in blood volume and blood oxygenation supporting neural activity [4]. Electroencephalographic technique is sensitive to variations in electrical potential in brain tissue caused by neuron depolarization and repolarization [2]. These methods reflect different aspects of the underlying physiological mechanisms, which include both vascular metabolic and electrophysiological processes mutually interacting in the visual cortex during visual stimulation. B. Near infrared spectroscopy Near Infrared Spectroscopy (NIRS) is a well-known optical technique first proposed by Franz Jöbsis in 1977 for invivo investigation of tissue oxygenation [6]. NIRS is based on the transparency of tissue to light in the spectral range between 700 and 1000 nm combined with the chromophore content of tissues: mainly water, deoxyhemoglobin and oxyhemoglobin. It has been well documented that NIRS is a useful technique for investigating cerebral hemodynamic changes in humans [4]. However, due to the relatively high scattering of the skull and white matter coefficients, it is difficult for NIR photons to penetrate the head for a depth greater than few centimetres. For this reason, NIRS is essentially limited to assessing cortical function [7,8]. The hemodynamic changes measured with NIRS are often called slow signals because they occur within seconds after brain activity begins. It has also known that NIRS has the capability to non-invasively measure neural activity [9-11]. Compared with the slow hemodynamic response, the neural signal (named "fast signal") occurs within milliseconds after stimulation. Researchers suggest that this signal originates from action potentials and the consequent swelling of the neural cells or from a brief period of anaerobic metabolism that somehow alters tissue transparency [12]. Page 2 of 15 (page number not for citation purposes) BioMedical Engineering OnLine 2007, 6:28 http://www.biomedical-engineering-online.com/content/6/1/28 Theory describing the migration of NIR photons through tissue is widely discussed in the literature [13-16]. The most accurate model is based on the photon diffusion equation that can be solved assuming that photons in tissue perform a random walk dominated by the high value of the scattering coefficient. This assumption is reasonable in near-infrared spectral region since the transport scattering coefficient μ's(λ) is about 50 times larger than the absorption coefficient μa(λ) in human tissues. In our previous paper [17], we proposed to combine this model with the knowledge of water concentration in human tissue. This approach allows the absolute variations of μa(λ) and μ's(λ) as well as changes in oxygenated and deoxygenated hemoglobin concentration (Δ[HbO2] and Δ[HHb], respectively) to be measured according to the following system of equations: where λ1, λ2 and λ3 represent the wavelengths of the photons used to explore the tissue, whereas αHbO2 and αHHb are the extinction coefficients of oxyhemoglobin and deoxyhemoglobin, respectively. The solution of this system is achieved by minimizing the least square errors. C. Electroencephalography ElectroEncephaloGraphy (EEG) is a common technique for recording the scalp surface electrical signal originating from electrical activity of the underlying brain structures. It includes complex information coming from all cerebral areas, and the signal amplitude, due to various somatosensorial stimulation, is maximal for electrodes located around the {associated/underlying} brain area. Essentially, the occipital area produces a response, as an EEG variation, for visual stimulation. Consequently, an international standardization for electrode placement has been introduced with the 10–20 System: the electrode positions O1 and O2 correspond to the left and right occipital hemisphere, respectively. A total visual field stimulation induces a similar response in both locations, which can be used to study visual functioning. However, when attempting to extract the visual response signal from the EEG, the very low signal-to-noise ratio imposes the need to acquire several EEG recordings to obtain reliable data and to process the pooled signals by means of averaging techniques to detect and evaluate specific evoked signals. Therefore, to improve the signal-tonoise ratio, the brain response must not be corrupted by other synchronous neural activities during boxcar averaging. When the sensory system involved in stimulation is the visual system, the response obtained from EEG through averaging is defined Visual Evoked Potential (VEP). The averaging algorithm implies the need for repetitive stimulation; depending on the stimulation intensity, normally 40 to 100 stimulations are needed to recover the VEP. The stimulation intensity is normally denoted according to some parameter of the stimulus applied; in our case we chose contrast, defined as , where L1 and L2 are the luminance values of adjacent areas of the stimulus. The VEP is interpreted as a function of time, VEP(t); its RMS value is defined as: where VEPk (k = 1,..., N) are the discrete values of the VEP after A/D conversion. VEPrms is the parameter strictly related to variation of the stimulus contrast, as we will see. D. System description 1. Block diagram and basic functions The block diagram of the developed measuring setup is shown in Figure 1. It consists of four main blocks: (a) optical unit, (b) electrical unit, (c) visual stimulation and electro-optical recording interface, and (d) control unit. 2. Optical unit In our previous paper [17], to overcome coupling between the scattering and absorption coefficients, we described a NIR-CWS instrument that allows quantitative assessments exploiting precise absorption measurements close to 975 nm, the absorption peak of water. Moreover, this system exploits an original detection scheme based on a time variant filter that approaches optimum shaping (in terms of signal to noise ratio) and has good rejection of photodetector offset, ambient light, and signal fluctuations due to probe movements. In this study we use the commercial release of our system produced by a spin-off company of our University: NIROX srl. To monitor two types of blood chromophores, we employed three laser diodes emitting different wavelengths in the near-infrared spectral region. The emitted radiation from each laser diode source is collimated into a single fiber. The back-diffused light is collected by a 3 mm diameter liquid optical guide (VIS-NIR 77634, Oriel Instruments, Germany). The receiver unit was designed to maximize the signal-tonoise ratio (SNR), reject continuous and alternate ambiΔ