Coregistration Errors Research Articles

Faster and improved 3-D head digitization in MEG using Kinect.

Accuracy in localizing the brain areas that generate neuromagnetic activity in magnetoencephalography (MEG) is dependent on properly co-registering MEG data to the participant's structural magnetic resonance image (MRI). Effective MEG-MRI co-registration is, in turn, dependent on how accurately we can digitize anatomical landmarks on the surface of the head. In this study, we compared the performance of three devices—Polhemus electromagnetic system, NextEngine laser scanner and Microsoft Kinect for Windows—for source localization accuracy and MEG-MRI co-registration. A calibrated phantom was used for verifying the source localization accuracy. The Kinect improved source localization accuracy over the Polhemus and the laser scanner by 2.23 mm (137%) and 0.81 mm (50%), respectively. MEG-MRI co-registration accuracy was verified on data from five healthy human participants, who received the digitization process using all three devices. The Kinect device captured approximately 2000 times more surface points than the Polhemus in one third of the time (1 min compared to 3 min) and thrice as many points as the NextEngine laser scanner. Following automated surface matching, the calculated mean MEG-MRI co-registration error for the Kinect was improved by 2.85 mm with respect to the Polhemus device, and equivalent to the laser scanner. Importantly, the Kinect device automatically aligns 20–30 images per second in real-time, reducing the limitations on participant head movement during digitization that are implicit in the NextEngine laser scan (~1 min). We conclude that the Kinect scanner is an effective device for head digitization in MEG, providing the necessary accuracy in source localization and MEG-MRI co-registration, while reducing digitization time.

Improving spatial localization in MEG inverse imaging by leveraging intersubject anatomical differences.

Modern neuroimaging techniques enable non-invasive observation of ongoing neural processing, with magnetoencephalography (MEG) in particular providing direct measurement of neural activity with millisecond time resolution. However, accurately mapping measured MEG sensor readings onto the underlying source neural structures remains an active area of research. This so-called “inverse problem” is ill posed, and poses a challenge for source estimation that is often cited as a drawback limiting MEG data interpretation. However, anatomically constrained MEG localization estimates may be more accurate than commonly believed. Here we hypothesize that, by combining anatomically constrained inverse estimates across subjects, the spatial uncertainty of MEG source localization can be mitigated. Specifically, we argue that differences in subject brain geometry yield differences in point-spread functions, resulting in improved spatial localization across subjects. To test this, we use standard methods to combine subject anatomical MRI scans with coregistration information to obtain an accurate forward (physical) solution, modeling the MEG sensor data resulting from brain activity originating from different cortical locations. Using a linear minimum-norm inverse to localize this brain activity, we demonstrate that a substantial increase in the spatial accuracy of MEG source localization can result from combining data from subjects with differing brain geometry. This improvement may be enabled by an increase in the amount of available spatial information in MEG data as measurements from different subjects are combined. This approach becomes more important in the face of practical issues of coregistration errors and potential noise sources, where we observe even larger improvements in localization when combining data across subjects. Finally, we use a simple auditory N100(m) localization task to show how this effect can influence localization using a recorded neural dataset.

A unifying probabilistic Bayesian approach to derive electron density from MRI for radiation therapy treatment planning

MRI significantly improves the accuracy and reliability of target delineation in radiation therapy for certain tumors due to its superior soft tissue contrast compared to CT. A treatment planning process with MRI as the sole imaging modality will eliminate systematic CT/MRI co-registration errors, reduce cost and radiation exposure, and simplify clinical workflow. However, MRI lacks the key electron density information necessary for accurate dose calculation and generating reference images for patient setup. The purpose of this work is to develop a unifying method to derive electron density from standard T1-weighted MRI. We propose to combine both intensity and geometry information into a unifying probabilistic Bayesian framework for electron density mapping. For each voxel, we compute two conditional probability density functions (PDFs) of electron density given its: (1) T1-weighted MRI intensity, and (2) geometry in a reference anatomy, obtained by deformable image registration between the MRI of the atlas and test patient. The two conditional PDFs containing intensity and geometry information are combined into a unifying posterior PDF, whose mean value corresponds to the optimal electron density value under the mean-square error criterion. We evaluated the algorithm’s accuracy of electron density mapping and its ability to detect bone in the head for eight patients, using an additional patient as the atlas or template. Mean absolute HU error between the estimated and true CT, as well as receiver operating characteristics for bone detection (HU > 200) were calculated. The performance was compared with a global intensity approach based on T1 and no density correction (set whole head to water). The proposed technique significantly reduced the errors in electron density estimation, with a mean absolute HU error of 126, compared with 139 for deformable registration (p = 2 × 10−4), 283 for the intensity approach (p = 2 × 10−6) and 282 without density correction (p = 5 × 10−6). For 90% sensitivity in bone detection, the proposed method achieved a specificity of 86%, compared with 80, 11 and 10% using deformable registration, intensity and without density correction, respectively. Notably, the Bayesian approach was more robust against anatomical differences between patients, with a specificity of 62% in the worst case (patient), compared to 30% specificity in registration-based approach. In conclusion, the proposed unifying Bayesian method provides accurate electron density estimation and bone detection from MRI of the head with highly heterogeneous anatomy.

A Unifying Method to Derive Electron Density From MRI for Radiation Therapy Treatment Planning

MRI has a number of advantages over CT as a primary modality for patient simulation and treatment planning in radiation therapy. Due to its superior soft tissue contrast, MRI significantly improves the accuracy and reliability of target delineation. An MRI based simulation will eliminate systematic image co-registration errors, reduce cost and simplify clinical workflow, with the additional benefit of zero ionizing radiation. However, one key bottleneck problem still remains, which is the lack of electron density information in MRI. The purpose of this work is to develop a reliable method to derive electron density from MRI. We adopt a probabilistic Bayesian approach for electron density mapping based on T1-weighted MRI. The probabilistic approach allows us to explicitly recognize the various sources of uncertainties (such as deformable registration and density mapping) and incorporate them into the analysis. We utilize a head and neck patient with co-registered CT/MRI as the atlas. For each voxel, we compute two conditional probabilities or likelihood functions: (1) electron density given its MR image intensity (T1 value), and (2) electron density given its geometry or position in a reference anatomy, obtained by deformable image registration between the MRI of the test patient and atlas. The two sources of information (image intensity and spatial location) are combined into a unifying posterior probability density function, using the Bayesian formalism. The mean value of the posterior or its mode (value that corresponds to the maximum probability) gives the electron density. To evaluate the proposed method, we calculated the mean absolute HU error between the estimated and true CT, as well as area under the ROC curve for bone detection (HU > 200). The performance was compared with a global intensity approach based on T1 value and no density correction (set whole head to water). The proposed technique significantly reduced the errors in electron density estimation, with a mean absolute HU error of 117, compared with 232 for the intensity approach and 273 without density correction. For bone detection, the proposed method had an accuracy of 87%, with a sensitivity of 90% and specificity of 90% (AUC = 94%). In comparison, the AUC for bone detection is only 70% and 50% using the intensity approach and without density correction, respectively. The proposed unifying method provides accurate electron density estimation and bone detection from MRI of the head with highly heterogeneous regions. This paves the way for accurate dose calculation and generating reference images for patient setup in MRI-based treatment planning.

Quantification of rat brain SPECT with 123I-ioflupane: evaluation of different reconstruction methods and image degradation compensations using Monte Carlo simulation

SPECT studies with 123I-ioflupane facilitate the diagnosis of Parkinson’s disease (PD). The effect on quantification of image degradations has been extensively evaluated in human studies but their impact on studies of experimental PD models is still unclear. The aim of this work was to assess the effect of compensating for the degrading phenomena on the quantification of small animal SPECT studies using 123I-ioflupane. This assessment enabled us to evaluate the feasibility of quantitatively detecting small pathological changes using different reconstruction methods and levels of compensation for the image degrading phenomena. Monte Carlo simulated studies of a rat phantom were reconstructed and quantified. Compensations for point spread function (PSF), scattering, attenuation and partial volume effect were progressively included in the quantification protocol. A linear relationship was found between calculated and simulated specific uptake ratio (SUR) in all cases. In order to significantly distinguish disease stages, noise-reduction during the reconstruction process was the most relevant factor, followed by PSF compensation. The smallest detectable SUR interval was determined by biological variability rather than by image degradations or coregistration errors. The quantification methods that gave the best results allowed us to distinguish PD stages with SUR values that are as close as 0.5 using groups of six rats to represent each stage.

Spectral light separator based on deep-subwavelength resonant apertures in a metallic film

We propose to funnel, select, and collect light spectrally by exploiting the unique properties of deep-subwavelength resonant apertures in a metallic film. In our approach, each aperture has an electromagnetic cross section that is much larger than its physical size while the frequency of the collected light is controlled by its height through the Fabry-Pérot resonance mechanism. The electromagnetic crosstalk between apertures remains low despite physical separations in the deep-subwavelength range. The resulting device enables an extremely efficient, subwavelength way to decompose light into its spectral components without the loss of photons and spatial coregistration errors. As a specific example, we show a subwavelength-size structure with three deep-subwavelength slits in a metallic film designed to operate in the mid-wave infrared range between 3 and 5.5 μm.

Improving the accuracy of local frequency estimation for interferometric synthetic aperture radar interferogram noise filtering considering large coregistration errors

This study deals with the problem of estimating the local frequencies for interferometric synthetic aperture radar (InSAR) phase image (i.e. interferogram) noise filtering considering large coregistration errors. The estimation of local frequencies is frequently applied to achieving high performance of interferometric phase noise filtering, which is a key step in InSAR processing procedures. Unfortunately, the generated interferograms suffer seriously from coregistration errors especially for complicated topographies, thus imposing strong limits on the accuracy of local frequencies estimation by the existing methods. Taking large coregistration errors into account, the proposed method makes full use of the neighbouring pixels to construct joint pixel vector and take the separation extent of the signal and noise subspaces as the criterion to accurately estimate the local frequencies so that the effects of the envelope misalignment can be mitigated. Theoretical analysis and the simulated data as well as the real ERS-1/2 data show that the presented method has the ability to provide accurate estimation of local frequencies even if the coregistration error reaches one pixel.

Robust Radial Velocity Estimation of Moving Targets Based on Adaptive Data Reconstruction and Subspace Projection Algorithm

In practice, inevitable image coregistration error and channel phase mismatch will significantly degrade the estimation performance of the target radial velocity in the ground moving target indication processing with multichannel synthetic aperture radar (SAR) systems. To overcome this problem, a new radial velocity estimation method using the subspace projection (SP) algorithm with adaptive data reconstruction is proposed in this letter. Based on the joint-pixel signal model, the Wiener weight vector is used to reconstruct the multichannel data vector of the pixel containing the moving target. Then, the SP algorithm is adopted to deal with the radial velocity estimation with the reconstructed single “snapshot” data. The validity and robustness are verified by both simulations and real SAR data experiments.

Robust differential interferometric synthetic aperture radar deformation phase estimation method based on covariance matrix information of multiple pixels

Differential interferometric synthetic aperture radar (D-InSAR) is a powerful technique to detect ground deformation, and the deformation estimation accuracy is directly affected by the D-InSAR phase accuracy. Coregistration of synthetic aperture radar (SAR) images and/or interferograms is one of the key steps in the D-InSAR flowchart, and the coregistration errors inevitable between SAR images will greatly decrease the D-InSAR phase estimation accuracy. In this study, the authors propose a robust approach to estimate the deformation phase by using the central pixel and its neighbouring pixels simultaneously, which is coupled with the signal subspace fitting or the covariance matrix fitting. The method can accomplish image auto-coregistration and suppress the D-InSAR phase noise together. Furthermore, compared with the required coregistration accuracy (no less than 1/10 pixel) for the conventional D-InSAR processing, the required image and/or interferogram coregistration error of the proposed method can reach 1 pixel, which greatly releases the district limitation on coregistration in D-InSAR processing. The performance is investigated by using simulated data and real ENVISAT ASAR data.

Consequences of EEG electrode position error on ultimate beamformer source reconstruction performance.

Inaccuracy of EEG electrode coordinates forms an error term in forward model generation and ultimate source reconstruction performance. This error arises from the combination of both intrinsic measurement noise of the digitization apparatus and manual coregistration error when selecting corresponding points on anatomical MRI volumes. A common assumption is that such an error would lead only to displacement of localized sources. Here, we measured electrode positions on a 3D-printed full-scale replica head, using three different techniques: a fringe projection 3D scanner, a novel “Flying Triangulation” 3D sensor, and a traditional electromagnetic digitizer. Using highly accurate fringe projection data as ground truth, the Flying Triangulation sensor had a mean error of 1.5 mm while the electromagnetic digitizer had a mean error of 6.8 mm. Then, again using the fringe projection as ground truth, individual EEG simulations were generated, with source locations across the brain space and a range of sensor noise levels. The simulated datasets were then processed using a beamformer in conjunction with the electrode coordinates registered with the Flying Triangulation and electromagnetic digitizer methods. The beamformer's output SNR was severely degraded with the digitizer-based positions but less severely with the Flying Triangulation coordinates. Therefore, the seemingly innocuous error in electrode registration may result in substantial degradation of beamformer performance, with output SNR penalties up to several decibels. In the case of low-SNR signals such as deeper brain structures or gamma band sources, this implies that sensor coregistration accuracy could make the difference between successful detection of such activity or complete failure to resolve the source.

3D change detection at street level using mobile laser scanning point clouds and terrestrial images

Automatic change detection and geo-database updating in the urban environment are difficult tasks. There has been much research on detecting changes with satellite and aerial images, but studies have rarely been performed at the street level, which is complex in its 3D geometry. Contemporary geo-databases include 3D street-level objects, which demand frequent data updating. Terrestrial images provides rich texture information for change detection, but the change detection with terrestrial images from different epochs sometimes faces problems with illumination changes, perspective distortions and unreliable 3D geometry caused by the lack of performance of automatic image matchers, while mobile laser scanning (MLS) data acquired from different epochs provides accurate 3D geometry for change detection, but is very expensive for periodical acquisition. This paper proposes a new method for change detection at street level by using combination of MLS point clouds and terrestrial images: the accurate but expensive MLS data acquired from an early epoch serves as the reference, and terrestrial images or photogrammetric images captured from an image-based mobile mapping system (MMS) at a later epoch are used to detect the geometrical changes between different epochs. The method will automatically mark the possible changes in each view, which provides a cost-efficient method for frequent data updating. The methodology is divided into several steps. In the first step, the point clouds are recorded by the MLS system and processed, with data cleaned and classified by semi-automatic means. In the second step, terrestrial images or mobile mapping images at a later epoch are taken and registered to the point cloud, and then point clouds are projected on each image by a weighted window based z-buffering method for view dependent 2D triangulation. In the next step, stereo pairs of the terrestrial images are rectified and re-projected between each other to check the geometrical consistency between point clouds and stereo images. Finally, an over-segmentation based graph cut optimization is carried out, taking into account the color, depth and class information to compute the changed area in the image space. The proposed method is invariant to light changes, robust to small co-registration errors between images and point clouds, and can be applied straightforwardly to 3D polyhedral models. This method can be used for 3D street data updating, city infrastructure management and damage monitoring in complex urban scenes.

High precision anatomy for MEG

Precise MEG estimates of neuronal current flow are undermined by uncertain knowledge of the head location with respect to the MEG sensors. This is either due to head movements within the scanning session or systematic errors in co-registration to anatomy. Here we show how such errors can be minimized using subject-specific head-casts produced using 3D printing technology. The casts fit the scalp of the subject internally and the inside of the MEG dewar externally, reducing within session and between session head movements. Systematic errors in matching to MRI coordinate system are also reduced through the use of MRI-visible fiducial markers placed on the same cast. Bootstrap estimates of absolute co-registration error were of the order of 1mm. Estimates of relative co-registration error were <1.5mm between sessions. We corroborated these scalp based estimates by looking at the MEG data recorded over a 6month period. We found that the between session sensor variability of the subject's evoked response was of the order of the within session noise, showing no appreciable noise due to between-session movement. Simulations suggest that the between-session sensor level amplitude SNR improved by a factor of 5 over conventional strategies. We show that at this level of coregistration accuracy there is strong evidence for anatomical models based on the individual rather than canonical anatomy; but that this advantage disappears for errors of greater than 5mm. This work paves the way for source reconstruction methods which can exploit very high SNR signals and accurate anatomical models; and also significantly increases the sensitivity of longitudinal studies with MEG.

Cone-beam Computed Tomography-guided Stereotactic Liver Punctures: A Phantom Study

Images from computed tomography (CT), combined with navigation systems, improve the outcomes of local thermal therapies that are dependent on accurate probe placement. Although the usage of CT is desired, its availability for time-consuming radiological interventions is limited. Alternatively, three-dimensional images from C-arm cone-beam CT (CBCT) can be used. The goal of this study was to evaluate the accuracy of navigated CBCT-guided needle punctures, controlled with CT scans. Five series of five navigated punctures were performed on a nonrigid phantom using a liver specific navigation system and CBCT volumetric dataset for planning and navigation. To mimic targets, five titanium screws were fixed to the phantom. Target positioning accuracy (TPECBCT) was computed from control CT scans and divided into lateral and longitudinal components. Additionally, CBCT-CT guidance accuracy was deducted by performing CBCT-to-CT image coregistration and measuring TPECBCT-CT from fused datasets. Image coregistration was evaluated using fiducial registration error (FRECBCT-CT) and target registration error (TRECBCT-CT). Positioning accuracies in lateral directions pertaining to CBCT (TPECBCT = 2.1 ± 1.0 mm) were found to be better to those achieved from previous study using CT (TPECT = 2.3 ± 1.3 mm). Image coregistration error was 0.3 ± 0.1 mm, resulting in an average TRE of 2.1 ± 0.7 mm (N = 5 targets) and average Euclidean TPECBCT-CT of 3.1 ± 1.3 mm. Stereotactic needle punctures might be planned and performed on volumetric CBCT images and controlled with multidetector CT with positioning accuracy higher or similar to those performed using CT scanners.

GeoCDX: An Automated Change Detection and Exploitation System for High-Resolution Satellite Imagery

We have developed a fully automated system for change detection of high-resolution satellite imagery. Our system, GeoCDX, is sensor-agnostic, resolution-independent and designed to process the very large volumes of data collected by modern high resolution panchromatic and multispectral imaging satellites. GeoCDX performs fully automated coregistration of imagery; extracts high-level features from the satellite imagery; performs change detection processing to pinpoint locations of change; clusters image tiles to group similar regions of change; and presents results in a variety of ways in an easy-to-use web application that facilitates online discovery, analysis, and dissemination of the change detection results. We applied GeoCDX to 4121 image pairs and successfully coregistered over 91% of the pairs covering a total area greater than 370 000 km2; GeoCDX decreased the average coregistration error from 9.6 ± 8.6 m to 1.8 ± 1.2 m. We show that for some pairs, GeoCDX provides up to a 50% increase in users' efficiency compared to manually performing change detection in common GIS software. Moreover, the change detection assessment performed using GeoCDX was on average four times more accurate compared to the manual approach in large part due to the use of our change intensity map that provides visual cues to the user during exploitation. Finally, change detection analysis using GeoCDX resulted in a missed detection rate of less than 2%.

Ground Moving Target Detection and Velocity Estimation Based on Spatial Multilook Processing for Multichannel Airborne SAR

A robust method is addressed here for ground moving target detection and velocity estimation. This method is proposed by forming multilook Doppler beam sharpening (DBS) images with different illuminated angles which are divided by slow time segment in the azimuth direction. Compared with only a single looking angle, the detection probability with multiple looking angles can be improved under the same condition. For the clutter rejection procedure, we describe an approach which utilizes the fast time information and therefore has the robustness to image coregisrtration error. After clutter rejection and target detection, a new series of single-look and multilook velocity estimation approaches are developed in the presence of channel mismatch and image coregistration error. The robustness and practicability of the proposed algorithm are verified by processing the measured data of the experiment with three-aperture airborne radar.

Accuracy assessment for mapping glacier flow velocity and detecting flow dynamics from ASTER satellite imagery: Tasman Glacier, New Zealand

Image matching via correlation of remotely sensed imagery is an increasingly popular technique for measuring glacier flow because it can provide spatially distributed measurements while mitigating many of the challenges accompanying in situ measurements of glacier velocity. Despite the increasing use of such techniques, and the requirement for well understood uncertainties where analysis of temporal variability is sought, studies that incorporate an assessment of accuracy based on in situ data remain rare. This study presents a rigorous methodology to quantify the uncertainties that apply to glacier flow velocities derived from optical satellite imagery. This method is applied on three successive ASTER images of Tasman Glacier, New Zealand. Evaluation of the quality of co-registration between image pair constituents allows unique uncertainties to be calculated for individual velocity measurements. These uncertainties have magnitudes that correspond to sub-pixel co-registration errors and account for anisotropic co-registration variance. Compared to estimated velocities, uncertainties on the order of 2–30% are achieved. Flow-fields for the 2009–2010 and 2010–2011 periods were derived using various spectral bands and validated with in situ data obtained by GPS survey over a nearly coincident period. This revealed the importance of considering the spectral characteristics of the target surface when selecting bands for image correlation, while validation demonstrated the success of the implemented methodology. Uncertainties derived for the flow-fields permitted statistical significance of velocity change between measurement periods to be assessed. Significant velocity changes, observed across the glacier surface, suggest that the Tasman Glacier behaves in a more dynamic way than previously recognised, highlighting the advantages of the methodology described here.

An estimation method for InSAR interferometric phase using correlation weight joint subspace projection

In this article, we propose a method to estimate the synthetic aperture radar interferometry (InSAR) interferometric phase based on the model of correlation weight joint pixel by using the joint subspace projection technique. In the method, the correlation weight joint data vector is constructed and the data vector can make the noise subspace dimension of the corresponding weight covariance matrix which is not affected by the coregistration error, thus avoiding the trouble of calculating the noise subspace dimension before estimating the InSAR interferometric phase. The method takes advantage of the coherence information of neighboring pixel pairs to auto-coregister the SAR images and employs the projection of the joint signal subspace onto the corresponding joint noise subspace to estimate the terrain interferometric phase. The method can auto-coregister the SAR images and reduce the interferometric phase noise simultaneously. Theoretical analysis and computer simulation results show that the method can provide accurate estimate of the interferometric phase (interferogram) even when the coregistration error reaches one pixel. The effectiveness of the method is verified via simulated data and real data.

Use of terrestrial laser scanning (TLS) for monitoring and modelling of geomorphic processes and phenomena at a small and medium spatial scale in Polar environment (Scott River — Spitsbergen)

The application of Terrestrial Laser Scanning (TLS) for precise modelling of land relief and quantitative estimation of spatial and temporal transformations can contribute to better understanding of catchment-forming processes. Experimental field measurements utilising the 3D laser scanning technology were carried out within the Scott River catchment located in the NW part of the Wedel Jarlsberg Land (Spitsbergen). The measurements concerned the glacier-free part of the Scott River valley floor with a length of 3.5km and width from 0.3 to 1.5km and were conducted with a state-of-the-art medium-range stationary laser scanner, a Leica Scan Station C10. A complex set of measurements of the valley floor were carried out from 86 measurement sites interrelated by the application of 82 common ‘target points’. During scanning, from 5 to 19 million measurements were performed at each of the sites, and a point-cloud constituting a ‘model space’ was obtained. By merging individual ‘model spaces’, a Digital Surface Model (DSM) of the Scott River valley was obtained, with a co-registration error not exceeding ±9mm. The accuracy of the model permitted precise measurements of dimensions of landforms of varied scales on the main valley floor and slopes and in selected sub-catchments. The analyses verified the efficiency of the measurement system in Polar meteorological conditions of Spitsbergen in mid-summer.

Clinical Application of In-Room Positron Emission Tomography for In Vivo Treatment Monitoring in Proton Radiation Therapy

The purpose of this study is to evaluate the potential of using in-room positron emission tomography (PET) for treatment verification in proton therapy and for deriving suitable PET scan times. Nine patients undergoing passive scattering proton therapy underwent scanning immediately after treatment with an in-room PET scanner. The scanner was positioned next to the treatment head after treatment. The Monte Carlo (MC) method was used to reproduce PET activities for each patient. To assess the proton beam range uncertainty, we designed a novel concept in which the measured PET activity surface distal to the target at the end of range was compared with MC predictions. The repositioning of patients for the PET scan took, on average, approximately 2 minutes. The PET images were reconstructed considering varying scan times to test the scan time dependency of the method. The measured PET images show overall good spatial correlations with MC predictions. Some discrepancies could be attributed to uncertainties in the local elemental composition and biological washout. For 8 patients treated with a single field, the average range differences between PET measurements and computed tomography (CT) image-based MC results were <5 mm (<3 mm for 6 of 8 patients) and root-mean-square deviations were 4 to 11 mm with PET-CT image co-registration errors of approximately 2 mm. Our results also show that a short-length PET scan of 5 minutes can yield results similar to those of a 20-minute PET scan. Our first clinical trials in 9 patients using an in-room PET system demonstrated its potential for invivo treatment monitoring in proton therapy. For a quantitative range prediction with arbitrary shape of target volume, we suggest using the distal PET activity surface.

Single subject fMRI test–retest reliability metrics and confounding factors

While the fMRI test-retest reliability has been mainly investigated from the point of view of group level studies, here we present analyses and results for single-subject test-retest reliability. One important aspect of group level reliability is that not only does it depend on between-session variance (test-retest), but also on between-subject variance. This has partly led to a debate regarding which reliability metric to use and how different sources of noise contribute to between-session variance. Focusing on single subject reliability allows considering between-session only. In this study, we measured test-retest reliability in four behavioural tasks (motor mapping, covert verb generation, overt word repetition, and a landmark identification task) to ensure generalisation of the results and at three levels of data processing (time-series correlation, t value variance, and overlap of thresholded maps) to understand how each step influences the other and how confounding factors influence reliability at each of these steps. The contributions of confounding factors (scanner noise, subject motion, and coregistration) were investigated using multiple regression and relative importance analyses at each step. Finally, to achieve a fuller picture of what constitutes a reliable task, we introduced a bootstrap technique of within- vs. between-subject variance. Our results show that (i) scanner noise and coregistration errors have little contribution to between-session variance (ii) subject motion (especially correlated with the stimuli) can have detrimental effects on reliability (iii) different tasks lead to different reliability results. This suggests that between-session variance in fMRI is mostly caused by the variability of underlying cognitive processes and motion correlated with the stimuli rather than technical limitations of data processing.