Ilkka Nissilä

Gauss-Newton method for image reconstruction in diffuse optical tomography

We present a regularized Gauss-Newton method for solving the inverse problem of parameter reconstruction from boundary data in frequency-domain diffuse optical tomography. To avoid the explicit formation and inversion of the Hessian which is often prohibitively expensive in terms of memory resources and runtime for large-scale problems, we propose to solve the normal equation at each Newton step by means of an iterative Krylov method, which accesses the Hessian only in the form of matrix-vector products. This allows us to represent the Hessian implicitly by the Jacobian and regularization term. Further we introduce transformation strategies for data and parameter space to improve the reconstruction performance. We present simultaneous reconstructions of absorption and scattering distributions using this method for a simulated test case and experimental phantom data.

Hemodynamic Responses to Speech and Music in Newborn Infants

We used near‐infrared spectroscopy (NIRS) to study responses to speech and music on the auditory cortices of 13 healthy full‐term newborn infants during natural sleep. The purpose of the study was to investigate the lateralization of speech and music responses at this stage of development. NIRS data was recorded from eight positions on both hemispheres simultaneously with electroencephalography, electrooculography, electrocardiography, pulse oximetry, and inclinometry. In 11 subjects, statistically significant (P < 0.02) oxygenated (HbO2) and total hemoglobin (HbT) responses were recorded. Both stimulus types elicited significant HbO2 and HbT responses on both hemispheres in five subjects. Six of the 11 subjects had positive HbO2 and HbT responses to both stimulus types, whereas one subject had negative responses. Mixed positive and negative responses were observed in four neonates. On both hemispheres, speech and music responses were significantly correlated (r = 0.64; P = 0.018 on the left hemisphere (LH) and r = 0.60; P = 0.029 on the right hemisphere (RH)). On the group level, the average response to the speech stimuli was statistically significantly greater than zero in the LH, whereas responses on the RH or to the music stimuli did not differ significantly from zero. This suggests a more coherent response to speech on the LH. However, significant differences in lateralization of the responses or mean response amplitudes of the two stimulus types were not observed on the group level. Hum Brain Mapp, 2010.

Coupling between somatosensory evoked potentials and hemodynamic response in the rat

We studied the relationship between somatosensory evoked potentials (SEP) recorded with scalp electroencephalography (EEG) and hemoglobin responses recorded non-invasively with diffuse optical imaging (DOI) during parametrically varied electrical forepaw stimulation in rats. Using these macroscopic techniques we verified that the hemodynamic response is not linearly coupled to the somatosensory evoked potentials, and that a power or threshold law best describes the coupling between SEP and the hemoglobin response, in agreement with the results of most invasive studies. We decompose the SEP response in three components (P1, N1, and P2) to determine which best predicts the hemoglobin response. We found that N1 and P2 predict the hemoglobin response significantly better than P1 and the input stimuli (S). Previous electrophysiology studies reported in the literature show that P1 originates in layer IV directly from thalamocortical afferents, while N1 and P2 originate in layers I and II and reflect the majority of local cortico-cortical interactions. Our results suggest that the evoked hemoglobin response is driven by the cortical synaptic activity and not by direct thalamic input. The N1 and P2 components, and not P1, need to be considered to correctly interpret neurovascular coupling.

Instrumentation and calibration methods for the multichannel measurement of phase and amplitude in optical tomography

In this article, we describe the multichannel implementation of an intensity modulated optical tomography system developed at Helsinki University of Technology. The system has two time-multiplexed wavelengths, 16 time-multiplexed source fibers and 16 parallel detection channels. The gain of the photomultiplier tubes (PMTs) is individually adjusted during the measurement sequence to increase the dynamic range of the system by 104. The PMT used has a high quantum efficiency in the near infrared (8% at 800nm), a fast settling time, and low hysteresis. The gain of the PMT is set so that the dc anode current is below 80nA, which allows the measurement of phase independently of the intensity. The system allows measurements of amplitude at detected intensities down to 1fW, which is sufficient for transmittance measurements of the female breast, the forearm, and the brain of early pre-term infants. The mean repeatability of phase and the logarithm of amplitude (lnA) at 100MHz were found to be 0.08° and 0.004, res...

Study of neurovascular coupling in humans via simultaneous magnetoencephalography and diffuse optical imaging acquisition

By combining diffuse optical imaging (DOI) and magnetoencephalography (MEG) we investigate neurovascular coupling non-invasively in human subjects using median-nerve stimulation. Previous fMRI studies have shown a habituation effect in the hemodynamic blood oxygen level-dependent (BOLD) response for stimulation periods longer than 2 s. With DOI and MEG we can test whether this effect in hemodynamic response can be accounted for by a habituation effect in the neural response. Our experimental results show that the habituation effect in the hemodynamic response is stronger than that in the earliest cortical neural response (N20). Using a linear convolution model to predict hemodynamic responses we found that including late neural components (> or = 30 ms) improves the prediction of the hemoglobin response. This finding suggests that in addition to the initial evoked-response deflections related to the talamic afferent input, later cortical activity is needed to predict the hemodynamic response.

Instrumentation for the accurate measurement of phase and amplitude in optical tomography

A single-channel prototype for a frequency-domain optical tomography system is presented. The two main goals in the design of the system were the measurement of phase with minimal systematic errors and a high enough signal-to-noise ratio to detect the small changes in the absorption of brain tissue during brain activity. Although the system inherently is an imaging system, the aspects of the system that relate to multichannel operation will be published separately, as this part of the system is not yet finished. The instrument is described in detail, including the radio-frequency system, the light detection system, and the light source. Factors that affect the accuracy of the measured phase include phase drift, radio-frequency coupling between the source and detector electronics, phase-amplitude cross talk, and others. To increase the range of intensities that can be measured, the gain of the detector is adjusted while keeping the mean anode current small compared with the quiescent current through the vo...

Approximation errors and model reduction in three-dimensional diffuse optical tomography

Model reduction is often required in diffuse optical tomography (DOT), typically because of limited available computation time or computer memory. In practice, this means that one is bound to use coarse mesh and truncated computation domain in the model for the forward problem. We apply the (Bayesian) approximation error model for the compensation of modeling errors caused by domain truncation and a coarse computation mesh in DOT. The approach is tested with a three-dimensional example using experimental data. The results show that when the approximation error model is employed, it is possible to use mesh densities and computation domains that would be unacceptable with a conventional measurement error model.

Image reconstruction in optical tomography in the presence of coupling errors.

Image reconstruction in optical tomography is a nonlinear and generally ill- posed inverse problem. Noise in the measured surface data can give rise to substantial artifacts in the recovered volume images of optical coefficients. Apart from random shot noise caused by the limited number of photons detected at the measurement site, another class of systematic noise is associated with losses specific to individual source and detector locations. A common cause for such losses in data acquisition systems based on fiber-optic light delivery is the imperfect coupling between the fiber tips and the skin of the patient because of air gaps or surface moisture. Thus the term coupling errors was coined for this type of data noise. However, source and detector specific errors can also occur in noncontact measurement systems not using fiber-optic delivery, for example, owing to local skin pigmentation, hair and hair follicles, or instrumentation calibration errors. Often it is not possible to quantify coupling effects in a way that allows us to remove them from the data or incorporate them into the light transport model. We present an alternative method of eliminating coupling errors by regarding the complex-valued coupling factors for each source and detector as unknowns in the reconstruction process and recovering them simultaneously with the images of absorption and scattering. Our method takes into account the possibility that coupling effects have an influence on both the amplitude and the phase shift of the measurements. Reconstructions from simulated and experimental phantom data are presented, which show that including the coupling coefficients in the reconstruction greatly improves the recovery of absorption and scattering images.

Modeling anisotropic light propagation in a realistic model of the human head

A Monte Carlo model capable of describing photon migration in arbitrary three-dimensional geometry with spatially varying optical properties and tissue anisotropy is presented. We use the model to explore the effects of anisotropy for optical measurements of the human head. An anisotropic diffusion equation that corresponds to our Monte Carlo model is derived, and a comparison between the Monte Carlo model and the diffusion equation solution with finite elements is given.

Significance of background optical properties, time-resolved information and optode arrangement in diffuse optical imaging of term neonates

The significance of accurate knowledge of background optical properties and time-resolved information in reconstructing images of hemodynamic changes in the neonatal brain from diffuse optical imaging data was studied using Monte Carlo (MC) simulation. A segmented anatomical magnetic resonance (MR) image and literature-derived optical properties for each tissue type were used to create a voxel-based anatomical model. Small absorbing perturbations were introduced into the anatomical model to simulate localized hemodynamic responses related to brain activation. Perturbation MC (pMC) was used as the primary method of image reconstruction. For comparison, reconstructions were also performed using the finite element method (FEM) to solve the diffusion approximation (DA) to the radiative transfer equation (RTE). The effect of optode layout was investigated using three different grids. Of the factors studied, the density of the optode grid was found to have the greatest effect on image quality. The use of time-resolved information significantly improved the spatial accuracy with all optode grids. Adequate knowledge and modeling of the optical properties of the background was found to significantly improve the spatial accuracy of the reconstructed images and make the recovery of contrast of absorption changes more consistent over simplified modeling. Localization accuracy of small perturbations was found to be 2-3 mm with accurate a priori knowledge of the background optical properties, when a grid with high optode density (>1 optode cm(-2)) was used.