Mutual information for automated unwarping of rat brain autoradiographs

Non-Equilibrium Enhancement of Classical Information Transmission

Information transmission plays a crucial role across various fields, including physics, engineering, biology, and society. The efficiency of this transmission is quantified by mutual information and its associated information capacity. While studies in closed systems have yielded significant progress, understanding the impact of non-equilibrium effects on open systems remains a challenge. These effects, characterized by the exchange of energy, information, and materials with the external environment, can influence both mutual information and information capacity. Here, we delve into this challenge by exploring non-equilibrium effects using the memoryless channel model, a cornerstone of information channel coding theories and methodology development. Our findings reveal that mutual information exhibits a convex relationship with non-equilibriumness, quantified by the non-equilibrium strength in transmission probabilities. Notably, channel information capacity is enhanced by non-equilibrium effects. Furthermore, we demonstrate that non-equilibrium thermodynamic cost, characterized by the entropy production rate, can actually improve both mutual information and information channel capacity, leading to a boost in overall information transmission efficiency. Our numerical results support our conclusions.

Rhesus Macaque Brain Atlas Regions Aligned to an MRI Template

To aid in the analysis of rhesus macaque brain images, we aligned digitized anatomical regions from the widely used atlas of Paxinos et al. to a published magnetic resonance imaging (MRI) template based on a large number of subjects. Digitally labelled atlas images were aligned to the template in 2D and then in 3D. The resulting grey matter regions appear qualitatively to be well registered to the template. To quantitatively validate the procedure, MR brain images of 20 rhesus macaques were aligned to the template along with regions drawn by hand in striatal and cortical areas in each subject's MRI. There was good geometric overlap between the hand drawn regions and the template regions. Positron emission tomography (PET) images of the same subjects showing uptake of a dopamine D₂ receptor ligand were aligned to the template space, and good agreement was found between tracer binding measures calculated using the hand drawn and template regions. In conclusion, an anatomically defined set of rhesus macaque brain regions has been aligned to an MRI template and has been validated for analysis of PET imaging in a subset of striatal and cortical areas. The entire set of over 200 regions is publicly available at https://www.nitrc.org/ . Graphical Abstract ᅟ.

Joint registration and synthesis using a probabilistic model for alignment of MRI and histological sections

Nonlinear registration of 2D histological sections with corresponding slices of MRI data is a critical step of 3D histology reconstruction algorithms. This registration is difficult due to the large differences in image contrast and resolution, as well as the complex nonrigid deformations and artefacts produced when sectioning the sample and mounting it on the glass slide. It has been shown in brain MRI registration that better spatial alignment across modalities can be obtained by synthesising one modality from the other and then using intra-modality registration metrics, rather than by using information theory based metrics to solve the problem directly. However, such an approach typically requires a database of aligned images from the two modalities, which is very difficult to obtain for histology and MRI.

Here, we overcome this limitation with a probabilistic method that simultaneously solves for deformable registration and synthesis directly on the target images, without requiring any training data. The method is based on a probabilistic model in which the MRI slice is assumed to be a contrast-warped, spatially deformed version of the histological section. We use approximate Bayesian inference to iteratively refine the probabilistic estimate of the synthesis and the registration, while accounting for each other’s uncertainty. Moreover, manually placed landmarks can be seamlessly integrated in the framework for increased performance and robustness.

Experiments on a synthetic dataset of MRI slices show that, compared with mutual information based registration, the proposed method makes it possible to use a much more flexible deformation model in the registration to improve its accuracy, without compromising robustness. Moreover, our framework also exploits information in manually placed landmarks more efficiently than mutual information: landmarks constrain the deformation field in both methods, but in our algorithm, it also has a positive effect on the synthesis – which further improves the registration. We also show results on two real, publicly available datasets: the Allen and BigBrain atlases. In both of them, the proposed method provides a clear improvement over mutual information based registration, both qualitatively (visual inspection) and quantitatively (registration error measured with pairs of manually annotated landmarks).

Automatic deformable registration of histological slides to μCT volume data

Localizing a histological section in the three-dimensional dataset of a different imaging modality is a challenging 2D-3D registration problem. In the literature, several approaches have been proposed to solve this problem; however, they cannot be considered as fully automatic. Recently, we developed an automatic algorithm that could successfully find the position of a histological section in a micro computed tomography (μCT) volume. For the majority of the datasets, the result of localization corresponded to the manual results. However, for some datasets, the matching μCT slice was off the ground-truth position. Furthermore, elastic distortions, due to histological preparation, could not be accounted for in this framework. In the current study, we introduce two optimization frameworks based on normalized mutual information, which enabled us to accurately register histology slides to volume data. The rigid approach allocated 81 % of histological sections with a median position error of 8.4 μm in jaw bone datasets, and the deformable approach improved registration by 33 μm with respect to the median distance error for four histological slides in the cerebellum dataset.

Transformation diffusion reconstruction of three-dimensional histology volumes from two-dimensional image stacks

•

A method for 3D reconstruction of serial 2D histology image stacks is proposed.

•

Pre-alignment to an external pre-cut reference (blockface) prevents shape artifacts.

•

Formulated as diffusion of transformations from each slice to its neighbors.

•

Registrations replaced by much faster transformation operations.

Traditional histology is the gold standard for tissue studies, but it is intrinsically reliant on two-dimensional (2D) images. Study of volumetric tissue samples such as whole hearts produces a stack of misaligned and distorted 2D images that need to be reconstructed to recover a congruent volume with the original sample's shape. In this paper, we develop a mathematical framework called Transformation Diffusion (TD) for stack alignment refinement as a solution to the heat diffusion equation. This general framework does not require contour segmentation, is independent of the registration method used, and is trivially parallelizable. After the first stack sweep, we also replace registration operations by operations in the space of transformations, several orders of magnitude faster and less memory-consuming. Implementing TD with operations in the space of transformations produces our Transformation Diffusion Reconstruction (TDR) algorithm, applicable to general transformations that are closed under inversion and composition. In particular, we provide formulas for translation and affine transformations. We also propose an Approximated TDR (ATDR) algorithm that extends the same principles to tensor-product B-spline transformations. Using TDR and ATDR, we reconstruct a full mouse heart at pixel size 0.92 µm × 0.92 µm, cut 10 µm thick, spaced 20 µm (84G). Our algorithms employ only local information from transformations between neighboring slices, but the TD framework allows theoretical analysis of the refinement as applying a global Gaussian low-pass filter to the unknown stack misalignments. We also show that reconstruction without an external reference produces large shape artifacts in a cardiac specimen while still optimizing slice-to-slice alignment. To overcome this problem, we use a pre-cutting blockface imaging process previously developed by our group that takes advantage of Brewster's angle and a polarizer to capture the outline of only the topmost layer of wax in the block containing embedded tissue for histological sectioning.

Patch-Based Nonlinear Image Registration for Gigapixel Whole Slide Images

OBJECTIVE

Image registration of whole slide histology images allows the fusion of fine-grained information-like different immunohistochemical stains-from neighboring tissue slides. Traditionally, pathologists fuse this information by looking subsequently at one slide at a time. If the slides are digitized and accurately aligned at cell level, automatic analysis can be used to ease the pathologist's work. However, the size of those images exceeds the memory capacity of regular computers.

METHODS

We address the challenge to combine a global motion model that takes the physical cutting process of the tissue into account with image data that is not simultaneously globally available. Typical approaches either reduce the amount of data to be processed or partition the data into smaller chunks to be processed separately. Our novel method first registers the complete images on a low resolution with a nonlinear deformation model and later refines this result on patches by using a second nonlinear registration on each patch. Finally, the deformations computed on all patches are combined by interpolation to form one globally smooth nonlinear deformation. The NGF distance measure is used to handle multistain images.

RESULTS

The method is applied to ten whole slide image pairs of human lung cancer data. The alignment of 85 corresponding structures is measured by comparing manual segmentations from neighboring slides. Their offset improves significantly, by at least 15%, compared to the low-resolution nonlinear registration.

CONCLUSION/SIGNIFICANCE

The proposed method significantly improves the accuracy of multistain registration which allows us to compare different antibodies at cell level.

Diffusion tensor imaging of the sural nerve in normal controls

OBJECTIVE

To develop a diffusion tensor imaging (DTI) protocol for assessing the sural nerve in healthy subjects.

METHODS

Sural nerves in 25 controls were imaged using DTI at 3T with 6, 15, and 32 gradient directions. Fractional anisotropy (FA) and apparent diffusion coefficient (ADC) were computed from nerve regions of interest co-registered with T2-weighted images.

RESULTS

Coronal images with 0.5(RL)× 2.0(FH)× 0.5(AP)mm(3) resolution successfully localized the sural nerve. FA maps showed less variability with 32 directions (0.559 ± 0.071) compared to 15(0.590 ± 0.080) and 6(0.659 ± 0.109).

CONCLUSIONS

Our DTI protocol was effective in imaging sural nerves in controls to establish normative FA/ADC, with potential to be used non-invasively in diseased nerves of patients.

3D prostate histology reconstruction: an evaluation of image-based and fiducial-based algorithms

PURPOSE

Evaluation of in vivo prostate imaging modalities for determining the spatial distribution and aggressiveness of prostate cancer ideally requires accurate registration of images to an accepted reference standard, such as histopathological examination of radical prostatectomy specimens. Three-dimensional (3D) reconstruction of prostate histology facilitates these registration-based evaluations by reintroducing 3D spatial information lost during histology processing. Because the reconstruction accuracy may constrain the clinical questions that can be answered with these data, it is important to assess the tradeoffs between minimally disruptive methods based on intrinsic image information and potentially more robust methods based on extrinsic fiducial markers.

METHODS

Ex vivo magnetic resonance (MR) images and digitized whole-mount histology images from 12 radical prostatectomy specimens were used to evaluate four 3D histology reconstruction algorithms. 3D reconstructions were computed by registering each histology image to the corresponding ex vivo MR image using one of two similarity metrics (mutual information or fiducial registration error) and one of two search domains (affine transformations or a constrained subset thereof). The algorithms were evaluated for accuracy using the mean target registration error (TRE) computed from homologous intrinsic point landmarks (3-16 per histology section; 232 total) identified on histology and MR images, and for the sensitivity of TRE to rotational, translational, and scaling initialization errors.

RESULTS

The algorithms using fiducial registration error and mutual information had mean ± standard deviation TREs of 0.7 ± 0.4 and 1.2 ± 0.7 mm, respectively, and one algorithm using fiducial registration error and affine transforms had negligible sensitivities to initialization errors. The postoptimization values of the mutual information-based metric showed evidence of errors due to both the optimizer and the similarity metric, and variation of parameters of the mutual information-based metric did not improve its performance.

CONCLUSIONS

The extrinsic fiducial-based algorithm had lower mean TRE and lower sensitivity to initialization than the intrinsic intensity-based algorithm using mutual information. A model relating statistical power to registration error for certain imaging validation study designs estimated that a reconstruction algorithm with a mean TRE of 0.7 mm would require 27% fewer subjects than the method used to initialize the algorithms (mean TRE 1.3 ± 0.7 mm), suggesting the choice of reconstruction technique can have a substantial impact on the design of imaging validation studies, and on their overall cost.

Integration of ultra-high field MRI and histology for connectome based research of brain disorders

Ultra-high field magnetic resonance imaging (MRI) became increasingly relevant for in vivo neuroscientific research because of improved spatial resolutions. However, this is still the unchallenged domain of histological studies, which long played an important role in the investigation of neuropsychiatric disorders. While the field of biological psychiatry strongly advanced on macroscopic levels, current developments are rediscovering the richness of immunohistological information when attempting a multi-level systematic approach to brain function and dysfunction. For most studies, histology sections lost information on three-dimensional reconstructions. Translating histological sections to 3D-volumes would thus not only allow for multi-stain and multi-subject alignment in post mortem data, but also provide a crucial step in big data initiatives involving the network analyses currently performed with in vivo MRI. We therefore investigated potential pitfalls during integration of MR and histological information where no additional blockface information is available. We demonstrated that strengths and requirements from both methods can be effectively combined at a spatial resolution of 200 μm. However, the success of this approach is heavily dependent on choices of hardware, sequence and reconstruction. We provide a fully automated pipeline that optimizes histological 3D reconstructions, providing a potentially powerful solution not only for primary human post mortem research institutions in neuropsychiatric research, but also to help alleviate the massive workloads in neuroanatomical atlas initiatives. We further demonstrate (for the first time) the feasibility and quality of ultra-high spatial resolution (150 μm isotopic) imaging of the entire human brain MRI at 7T, offering new opportunities for analyses on MR-derived information.

Surface fluid registration of conformal representation: application to detect disease burden and genetic influence on hippocampus

In this paper, we develop a new automated surface registration system based on surface conformal parameterization by holomorphic 1-forms, inverse consistent surface fluid registration, and multivariate tensor-based morphometry (mTBM). First, we conformally map a surface onto a planar rectangle space with holomorphic 1-forms. Second, we compute surface conformal representation by combining its local conformal factor and mean curvature and linearly scale the dynamic range of the conformal representation to form the feature image of the surface. Third, we align the feature image with a chosen template image via the fluid image registration algorithm, which has been extended into the curvilinear coordinates to adjust for the distortion introduced by surface parameterization. The inverse consistent image registration algorithm is also incorporated in the system to jointly estimate the forward and inverse transformations between the study and template images. This alignment induces a corresponding deformation on the surface. We tested the system on Alzheimer's Disease Neuroimaging Initiative (ADNI) baseline dataset to study AD symptoms on hippocampus. In our system, by modeling a hippocampus as a 3D parametric surface, we nonlinearly registered each surface with a selected template surface. Then we used mTBM to analyze the morphometry difference between diagnostic groups. Experimental results show that the new system has better performance than two publicly available subcortical surface registration tools: FIRST and SPHARM. We also analyzed the genetic influence of the Apolipoprotein E[element of]4 allele (ApoE4), which is considered as the most prevalent risk factor for AD. Our work successfully detected statistically significant difference between ApoE4 carriers and non-carriers in both patients of mild cognitive impairment (MCI) and healthy control subjects. The results show evidence that the ApoE genotype may be associated with accelerated brain atrophy so that our work provides a new MRI analysis tool that may help presymptomatic AD research.

Challenges in accurate registration of 3-D medical imaging and histopathology in primary prostate cancer

Due to poor correlation between slice thickness and orientation, verification of medical imaging results by histology is difficult. Often validation of imaging findings of lesions suspicious for prostate cancer is driven by a subjective, visual approach to correlate in vivo images with histopathology. We describe fallacious assumptions in the correlation of imaging findings with pathology and identify the lack of accurate registration as a major obstacle in the validation of PET and PET/CT imaging in primary prostate cancer. Specific registration techniques that facilitate the most difficult part of the registration process--the mapping of pathology onto high-resolution imaging, preferably aided by the ex vivo prostate specimen--are discussed.

Combined MR, fluorescence and histology imaging strategy in a human breast tumor xenograft model

Applications of molecular imaging in cancer and other diseases frequently require the combination of in vivo imaging modalities, such as MR and optical imaging, with ex vivo optical, fluorescence, histology and immunohistochemical imaging to investigate and relate molecular and biological processes to imaging parameters within the same region of interest. We have developed a multimodal image reconstruction and fusion framework that accurately combines in vivo MRI and MRSI, ex vivo brightfield and fluorescence microscopic imaging and ex vivo histology imaging. Ex vivo brightfield microscopic imaging was used as an intermediate modality to facilitate the ultimate link between ex vivo histology and in vivo MRI/MRSI. Tissue sectioning necessary for optical and histology imaging required the generation of a three-dimensional reconstruction module for two-dimensional ex vivo optical and histology imaging data. We developed an external fiducial marker-based three-dimensional reconstruction method, which was able to fuse optical brightfield and fluorescence with histology imaging data. The registration of the three-dimensional tumor shape was pursued to combine in vivo MRI/MRSI and ex vivo optical brightfield and fluorescence imaging data. This registration strategy was applied to in vivo MRI/MRSI, ex vivo optical brightfield/fluorescence and histology imaging datasets obtained from human breast tumor models. Three-dimensional human breast tumor datasets were successfully reconstructed and fused with this platform.

MRI-guided volume reconstruction of mouse brain from histological sections

A method is presented for three-dimensional reconstruction of the mouse brain from histological sections with the guidance of magnetic resonance images (MRI). A major focus of the method is dealing with sections in which anatomical structures have been separated or distorted as a result of histological processing. Although histology has superb resolution with the ability to discriminate cell types and anatomical structures, misalignment between sections and distortion within sections renders 3D reconstruction of the histology volume simply by stacking 2D sections inadequate. In contrast, MRI preserves the spatial and geometric information about structures at a cost of cellular detail. To utilize the information from MRI in reconstructing volumetric histological data, we developed a procedure consisting of a series of segmentation and registration operations. The method is iterative and first identifies the corresponding MRI slices for each histological section. Piecewise rigid registration is then employed to deal with tissue distortion caused by histological processing. Quantitative validation of the method's accuracy was performed on four reconstructed mouse brains by comparing a set of manually selected anatomical landmarks on pairs of MRI and histological volumes. The procedure is highly automated and amenable to high throughput.

Landmark optimization using local curvature for point-based nonlinear rodent brain image registration

Purpose. To develop a technique to automate landmark selection for point-based interpolating transformations for nonlinear medical image registration. Materials and Methods. Interpolating transformations were calculated from homologous point landmarks on the source (image to be transformed) and target (reference image). Point landmarks are placed at regular intervals on contours of anatomical features, and their positions are optimized along the contour surface by a function composed of curvature similarity and displacements of the homologous landmarks. The method was evaluated in two cases (n = 5 each). In one, MRI was registered to histological sections; in the second, geometric distortions in EPI MRI were corrected. Normalized mutual information and target registration error were calculated to compare the registration accuracy of the automatically and manually generated landmarks. Results. Statistical analyses demonstrated significant improvement (P < 0.05) in registration accuracy by landmark optimization in most data sets and trends towards improvement (P < 0.1) in others as compared to manual landmark selection.

Smoothness-guided 3-D reconstruction of 2-D histological images

This paper tackles two problems: (1) the reconstruction of 3-D volumes from 2-D post-mortem slices (e.g., histology, autoradiography, immunohistochemistry) in the absence of external reference, and (2) the quantitative evaluation of the 3-D reconstruction. We note that the quality of a reconstructed volume is usually assessed by considering the smoothness of some reconstructed structures of interest (e.g., the gray-white matter surfaces in brain images). Here we propose to use smoothness as a means to drive the reconstruction process itself. From a pair-wise rigid reconstruction of the 2-D slices, we first extract the boundaries of structures of interest. Those are then smoothed with a min-max curvature flow confined to the 2-D planes in which the slices lie. Finally, for each slice, we estimate a linear or flexible transformation from the sparse displacement field computed from the flow, which we apply to the original 2-D slices to obtain a smooth volume. In addition, we present a co-occurrence matrix-based technique to quantify the smoothness of reconstructed volumes. We discuss and validate the application of both our reconstruction approach and the smoothness measure on synthetic examples as well as real histological data.

Large Deformation Diffeomorphic Metric Mapping Registration of Reconstructed 3D Histological Section Images and in vivo MR Images

Our current understanding of neuroanatomical abnormalities in neuropsychiatric diseases is based largely on magnetic resonance imaging (MRI) and post mortem histological analyses of the brain. Further advances in elucidating altered brain structure in these human conditions might emerge from combining MRI and histological methods. We propose a multistage method for registering 3D volumes reconstructed from histological sections to corresponding in vivo MRI volumes from the same subjects: (1) manual segmentation of white matter (WM), gray matter (GM) and cerebrospinal fluid (CSF) compartments in histological sections, (2) alignment of consecutive histological sections using 2D rigid transformation to construct a 3D histological image volume from the aligned sections, (3) registration of reconstructed 3D histological volumes to the corresponding 3D MRI volumes using 3D affine transformation, (4) intensity normalization of images via histogram matching, and (5) registration of the volumes via intensity based large deformation diffeomorphic metric (LDDMM) image matching algorithm. Here we demonstrate the utility of our method in the transfer of cytoarchitectonic information from histological sections to identify regions of interest in MRI scans of nine adult macaque brains for morphometric analyses. LDDMM improved the accuracy of the registration via decreased distances between GM/CSF surfaces after LDDMM (0.39 ± 0.13 mm) compared to distances after affine registration (0.76 ± 0.41 mm). Similarly, WM/GM distances decreased to 0.28 ± 0.16 mm after LDDMM compared to 0.54 ± 0.39 mm after affine registration. The multistage registration method may find broad application for mapping histologically based information, for example, receptor distributions, gene expression, onto MRI volumes.

Towards ultra-high resolution fibre tract mapping of the human brain - registration of polarised light images and reorientation of fibre vectors

Polarised light imaging (PLI) utilises the birefringence of the myelin sheaths in order to visualise the orientation of nerve fibres in microtome sections of adult human post-mortem brains at ultra-high spatial resolution. The preparation of post-mortem brains for PLI involves fixation, freezing and cutting into 100-μm-thick sections. Hence, geometrical distortions of histological sections are inevitable and have to be removed for 3D reconstruction and subsequent fibre tracking. We here present a processing pipeline for 3D reconstruction of these sections using PLI derived multimodal images of post-mortem brains. Blockface images of the brains were obtained during cutting; they serve as reference data for alignment and elimination of distortion artefacts. In addition to the spatial image transformation, fibre orientation vectors were reoriented using the transformation fields, which consider both affine and subsequent non-linear registration. The application of this registration and reorientation approach results in a smooth fibre vector field, which reflects brain morphology. PLI combined with 3D reconstruction and fibre tracking is a powerful tool for human brain mapping. It can also serve as an independent method for evaluating in vivo fibre tractography.

Evaluation of treatment-associated inflammatory response on diffusion-weighted magnetic resonance imaging and 2-[18F]-fluoro-2-deoxy-D-glucose-positron emission tomography imaging biomarkers

PURPOSE

Functional imaging biomarkers of cancer treatment response offer the potential for early determination of outcome through the assessment of biochemical, physiologic, and microenvironmental readouts. Cell death may result in an immunologic response, thus complicating the interpretation of biomarker readouts. This study evaluated the temporal effect of treatment-associated inflammatory activity on diffusion magnetic resonance imaging and 2-[(18)F]-fluoro-2-deoxy-D-glucose-positron emission tomography imaging (FDG-PET) biomarkers to delineate the effects of the inflammatory response on imaging readouts.

EXPERIMENTAL DESIGN

Rats with intracerebral 9L gliosarcomas were separated into four groups consisting of control, an immunosuppressive agent dexamethasone (Dex), 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU), and BCNU+Dex. Animals were imaged using diffusion-weighted magnetic resonance imaging and FDG-PET at 0, 3, and 7 days posttreatment.

RESULTS

In the BCNU- and BCNU+Dex-treated animal groups, diffusion values increased progressively over the 7-day study period to approximately 23% over baseline. The FDG percentage change of standard uptake value decreased at day 3 (-30.9%) but increased over baseline levels at day 7 (+20.1%). FDG-PET of BCNU+Dex-treated animals were found to have percentage of standard uptake value reductions of -31.4% and -24.7% at days 3 and 7, respectively, following treatment. Activated macrophages were observed on day 7 in the BCNU treatment group with much fewer found in the BCNU+Dex group.

CONCLUSIONS

Results revealed that treatment-associated inflammatory response following tumor therapy resulted in the accentuation of tumor diffusion response along with a corresponding increase in tumor FDG uptake due to the presence of glucose-consuming activated macrophages. The dynamics and magnitude of potential inflammatory response should be considered when interpreting imaging biomarker results.

Smooth 3-D reconstruction for 2-D histological images

We present an image driven approach to the reconstruction of 3-D volumes from stacks of 2-D post-mortem sections (histology, cryoimaging, autoradiography or immunohistochemistry) in the absence of any external information. We note that a desirable quality of the reconstructed volume is the smoothness of its notable structures (e.g. the gray/white matter surfaces in brain images). Here we propose to use smoothness as a means to drive the reconstruction process itself. From an initial rigid pair-wise reconstruction of the input 2-D sections, we extract the boundaries of structures of interest. Those are then evolved under a mean curvature flow modified to constrain the flow within 2-D planes. Sparse displacement fields are then computed, independently for each slice, from the resulting flow. A variety of transformations, from globally rigid to arbitrarily flexible ones, can then be estimated from those fields and applied to the individual input 2-D sections to form a smooth volume. We detail our method and discuss preliminary results on both real histological data and synthetic examples.

Evaluation of registration strategies for multi-modality images of rat brain slices

In neuroscience, small-animal studies frequently involve dealing with series of images from multiple modalities such as histology and autoradiography. The consistent and bias-free restacking of multi-modality image series is obligatory as a starting point for subsequent non-rigid registration procedures and for quantitative comparisons with positron emission tomography (PET) and other in vivo data. Up to now, consistency between 2D slices without cross validation using an inherent 3D modality is frequently presumed to be close to the true morphology due to the smooth appearance of the contours of anatomical structures. However, in multi-modality stacks consistency is difficult to assess. In this work, consistency is defined in terms of smoothness of neighboring slices within a single modality and between different modalities. Registration bias denotes the distortion of the registered stack in comparison to the true 3D morphology and shape. Based on these metrics, different restacking strategies of multi-modality rat brain slices are experimentally evaluated. Experiments based on MRI-simulated and real dual-tracer autoradiograms reveal a clear bias of the restacked volume despite quantitatively high consistency and qualitatively smooth brain structures. However, different registration strategies yield different inter-consistency metrics. If no genuine 3D modality is available, the use of the so-called SOP (slice-order preferred) or MOSOP (modality-and-slice-order preferred) strategy is recommended.

Feature-based rectal contour propagation from planning CT to cone beam CT

The purpose of this work is to develop a novel feature-based registration strategy to automatically map the rectal contours from planning computed tomography (CT) (pCT) to cone beam CT (CBCT). The rectal contours were manually outlined on the pCT. A narrow band with the outlined contour as its interior surface was then constructed, so that we can exclude the volume inside the rectum in the registration process. The corresponding contour in the CBCT was found by using a feature-based registration algorithm, which consists of two steps: (1) automatically searching for control points in the pCT and CBCT based on the features of the surrounding tissue and matching the homologous control points using the scale invariance feature transformation; and (2) using the control points for a thin plate spline transformation to warp the narrow band and mapping the corresponding contours from pCT to CBCT. The proposed contour propagation technique is applied to digital phantoms and clinical cases and, in all cases, the contour mapping results are found to be clinically acceptable. For clinical cases, the method yielded satisfactory results even when there were significant rectal content changes between the pCT and CBCT scans. As a consequence, the accordance between the rectal volumes after deformable registration and the manually segmented rectum was found to be more than 90%. The proposed technique provides a powerful tool for adaptive radiotherapy of prostate, rectal, and gynecological cancers in the future.

Information entropy measure for evaluation of image quality

This paper presents a simple and straightforward method for synthetically evaluating digital radiographic images by a single parameter in terms of transmitted information (TI). The features of our proposed method are (1) simplicity of computation, (2) simplicity of experimentation, and (3) combined assessment of image noise and resolution (blur). Two acrylic step wedges with 0-1-2-3-4-5 and 0-2-4-6-8-10 mm in thickness were used as phantoms for experiments. In the present study, three experiments were conducted. First, to investigate the relation between the value of TI and image noise, various radiation doses by changing exposure time were employed. Second, we examined the relation between the value of TI and image blurring by shifting the phantoms away from the center of the X-ray beam area toward the cathode end when imaging was performed. Third, we analyzed the combined effect of deteriorated blur and noise on the images by employing three smoothing filters. Experimental results show that the amount of TI is closely related to both image noise and image blurring. The results demonstrate the usefulness of our method for evaluation of physical image quality in medical imaging.

Formulation of current density weighted indices for correspondence between functional MRI and electrocortical stimulation maps

OBJECTIVE

Accurate localization of functionally significant brain regions reduces risks of post-operative neurological deficits. The gold standard for presurgical brain mapping is subdural electrocortical stimulation (ECS), which is an open-cranium surgical procedure. Functional MRI (fMRI) may be a noninvasive alternative if it can be shown that fMRI and ECS maps are spatially consistent. We formulate new 3D current density weighted ECS-fMRI correspondence indices and illustrate their use on human data.

METHODS

Current density maps were computed for simulated and human datasets by solving the electrostatic Laplace equation. The proposed indices were characterized and compared with fixed radii and Euclidean distance indices.

RESULTS

Results from simulated datasets showed that the proposed indices quantify correspondence between fMRI and the ECS truth predictably, and provide conspicuous sensitivity increase from fixed radii indices, whereas Euclidean distances may not be suitable measures of the correspondence.

CONCLUSIONS

The proposed indices reflect contextual information from surrounding electrodes and may be physiologically more meaningful in evaluating ECS-fMRI correspondence.

SIGNIFICANCE

To identify safe limits of resection, an ECS map requires placement of electrodes on a patient's brain. Our proposed indices accurately quantify ECS-fMRI correspondence and may be used to evaluate fMRI as a noninvasive alternative for defining resection limits.

Registration methodology for histological sections and in vivo imaging of human prostate

RATIONALE AND OBJECTIVES

Registration enables quantitative spatial correlation of features from different imaging modalities. Our objective is to register in vivo imaging with histologic sections of the human prostate so that histologic truth can be correlated with in vivo imaging features.

MATERIALS AND METHODS

In vivo imaging of the prostate included T2-weighted anatomic and diffusion weighted 3-T magnetic resonance imaging (MRI) as well as 11C-choline positron emission tomography (PET). In addition, ex vivo 3-T MRI of the prostate specimen, histology, and associated block face photos of the prostate specimen were obtained. A standard registration method based on mutual information (MI) and thin-plate spline (TPS) was applied. Registration among in vivo imaging modalities is well established; however, accurate registration involving histology is difficult. Our approach breaks up the difficult direct registration of histology and in vivo imaging into achievable subregistration tasks involving intermediate ex vivo modalities like block face photography and specimen MRI. Results of subregistration tasks are combined to compute the intended, final registration between in vivo imaging and histology.

RESULTS

The methodology was applied to two patients and found to be clinically feasible. Overall registered anatomic MRI, diffusion MRI, and 11C-choline PET aligned well with histology qualitatively for both patients. There is no ground truth of registration accuracy as the scans are real patient scans. An indirect validation of the registration accuracy has been proposed comparing tumor boundary markings found in diffusion MRI and histologic sections. Registration errors for two patients between diffusion MRI and histology were 3.74 and 2.26 mm.

CONCLUSION

This proof of concept paper demonstrates a method based on intrinsic image information content for successfully registering in vivo imaging of the human prostate with its post-resection histology, which does not require the use of extrinsic fiducial markers. The methodology successfully mapped histology onto the in vivo imaging space, allowing the observation of how well different in vivo imaging features correspond to histologic truth. The methodology is therefore the basis for a systematic comparison of in vivo imaging for staging of human prostate cancer.

Brain maps on the go: functional imaging during motor challenge in animals

Brain mapping in the freely moving animal is useful for studying motor circuits, not only because it avoids the potential confound of sedation or restraints, but because activated brain states may serve to accentuate differences that only manifest partially while a subject is in the resting state. Perfusion or metabolic mapping using autoradiography allows one to examine changes in brain function at the circuit level across the entire brain with a spatial resolution (approximately 100 micro) appropriate for the rat or mouse brain, and a temporal resolution (seconds-minutes) sufficient for capturing acute brain changes. Here we summarize the application of these methods to the functional brain mapping of behaviors involving locomotion of small animals, methods for the three-dimensional reconstruction of the brain from autoradiographic sections, voxel based analysis of the whole brain, and generation of maps of the flattened rat cortex. Application of these methods in animal models promises utility in improving our understanding of motor function in the normal brain, and of the effects of neuropathology and treatment interventions such as exercise have on the reorganization of motor circuits.

Ingress of blood-borne macrophages across the blood-brain barrier in murine HIV-1 encephalitis

Blood-borne macrophage ingress into brain in HIV-1 associated neurocognitive disorders governs the tempo of disease. We used superparamagnetic iron-oxide particles loaded into murine bone marrow-derived macrophages (BMM) injected intravenously into HIV-1 encephalitis mice to quantitatively assess BMM entry into diseased brain regions. Magnetic resonance imaging tests were validated by histological coregistration and enhanced image processing. The demonstration of robust BMM migration into areas of focal encephalitis provide 'proof of concept' for the use of MRI to monitor macrophage ingress into brain.

The Neuroterrain 3D Mouse Brain Atlas

A significant objective of neuroinformatics is the construction of tools to readily access, search, and analyze anatomical imagery. This goal can be subdivided into development of the necessary databases and of the computer vision tools for image analysis. When considering mesoscale images, the latter tools can be further divided into registration algorithms and anatomical models. The models are atlases that contain both bitmap images and templates of anatomical boundaries. We report here on construction of such a model for the C57BL/6J mouse. The intended purpose of this atlas is to aid in automated delineation of the Mouse Brain Library, a database of brain histological images of importance to neurogenetic research.

Brain spatial normalization

Neuroanatomical informatics, a subspecialty of neuroinformatics, focuses on technological solutions to neuroimage database access. Its current main goal is an image-based query system that is able to retrieve imagery based on anatomical location. Here, we describe a set of tools that collectively form such a solution for sectional material and that are available to investigators to use on their own data sets. The system accepts slide images as input and yields a matrix of transformation parameters that map each point on the input image to a standardized 3D brain atlas. In essence, this spatial normalization makes the atlas a spatial indexer from which queries can be issued simply by specifying a location on the reference atlas. Our objective here is to familiarize potential users of the system with the steps required of them as well as steps that take place behind the scene. We detail the capabilities and the limitations of the current implementation and briefly describe the enhancements planned for the near future.

Alignment of large image series using cubic B-splines tessellation: application to transmission electron microscopy data

3D reconstruction from serial 2D microscopy images depends on non-linear alignment of serial sections. For some structures, such as the neuronal circuitry of the brain, very large images at very high resolution are necessary to permit reconstruction. These very large images prevent the direct use of classical registration methods. We propose in this work a method to deal with the non-linear alignment of arbitrarily large 2D images using the finite support properties of cubic B-splines. After initial affine alignment, each large image is split into a grid of smaller overlapping sub-images, which are individually registered using cubic B-splines transformations. Inside the overlapping regions between neighboring sub-images, the coefficients of the knots controlling the B-splines deformations are blended, to create a virtual large grid of knots for the whole image. The sub-images are resampled individually, using the new coefficients, and assembled together into a final large aligned image. We evaluated the method on a series of large transmission electron microscopy images and our results indicate significant improvements compared to both manual and affine alignment.

Quantitative validation of voxel-wise statistical analyses of autoradiographic rat brain volumes: application to unilateral visual stimulation

PET scanners devoted to in vivo functional study have recently been developed, but autoradiography remains the reference technique for assessing cerebral glucose metabolism (CMRGlu) in rodents. Autoradiographs are conventionally subjected to region of interest (ROI) analysis, which is intrinsically hypothesis-driven and therefore not suitable for whole-brain investigation. Voxel-wise statistical methods of analysis have long been used to determine differences in brain activity during in vivo functional neuroimaging experiments. They have also recently been applied to 3D reconstructed autoradiographic volume images from rat brains. We present here a fully automated analysis for autoradiographic data combining (1) computerized procedures for the acquisition and 3D reconstruction of postmortem volume images and (2) spatial normalization followed by classical whole-brain voxel-wise statistical analysis. We also describe an additional procedure for characterizing functional differences between the right and left hemispheres of the brain. We compared two spatial normalization techniques and evaluated how the effect of choosing a particular normalization technique impacted on the statistical analysis. We also propose a small volume correction analysis to address the problem of multiple statistical comparisons. Lastly, we investigated the reliability of such analyses, by comparing their results qualitatively and quantitatively with those previously obtained with our semiautomated ROI-based analysis [Dubois, A., Dauguet, J., Herard, A.-S., Besret, L., Duchesnay, E., Frouin, V., Hantraye, P., Bonvento, G., Delzescaux, T., 2007. Automated three-dimensional analysis of histologic and autoradiographic rat brain sections: application to an activation study. J. Cereb. Blood Flow Metab. 27 (10), 1742-1755.]. Both voxel-wise statistical analyses led to the detection of consistent interhemispheric differences in CMRGlu. This work demonstrates the potential value and robustness of voxel-wise statistical methods for analyzing autoradiographic data sets.

Automated three-dimensional analysis of histological and autoradiographic rat brain sections: application to an activation study

Besides the newly developed positron emission tomography scanners (microPET) dedicated to the in vivo functional study of small animals, autoradiography remains the reference technique widely used for functional brain imaging and the gold standard for the validation of in vivo results. The analysis of autoradiographic data is classically achieved in two dimensions (2D) using a section-by-section approach, is often limited to few sections and the delineation of the regions of interest to be analysed is directly performed on autoradiographic sections. In addition, such approach of analysis does not accommodate the possible anatomical shifts linked to dissymmetry associated with the sectioning process. This classic analysis is time-consuming, operator-dependent and can therefore lead to non-objective and non-reproducible results. In this paper, we have developed an automated and generic toolbox for processing of autoradiographic and corresponding histological rat brain sections based on a three-step approach, which involves: (1) an optimized digitization dealing with hundreds of autoradiographic and histological sections; (2) a robust reconstruction of the volumes based on a reliable registration method; and (3) an original 3D-geometry-based approach to analysis of anatomical and functional post-mortem data. The integration of the toolbox under a unified environment (in-house software BrainVISA, http://brainvisa.info) with a graphic interface enabled a robust and operator-independent exploitation of the overall anatomical and functional information. We illustrated the substantial qualitative and quantitative benefits obtained by applying our methodology to an activation study (rats, n=5, under unilateral visual stimulation).

Statistical properties of Jacobian maps and the realization of unbiased large-deformation nonlinear image registration

Maps of local tissue compression or expansion are often computed by comparing magnetic resonance imaging (MRI) scans using nonlinear image registration. The resulting changes are commonly analyzed using tensor-based morphometry to make inferences about anatomical differences, often based on the Jacobian map, which estimates local tissue gain or loss. Here, we provide rigorous mathematical analyses of the Jacobian maps, and use themto motivate a new numerical method to construct unbiased nonlinear image registration. First, we argue that logarithmic transformation is crucial for analyzing Jacobian values representing morphometric differences. We then examine the statistical distributions of log-Jacobian maps by defining the Kullback-Leibler (KL) distance on material density functions arising in continuum-mechanical models. With this framework, unbiased image registration can be constructed by quantifying the symmetric KL-distance between the identity map and the resulting deformation. Implementation details, addressing the proposed unbiased registration as well as the minimization of symmetric image matching functionals, are then discussed and shown to be applicable to other registration methods, such as inverse consistent registration. In the results section, we test the proposed framework, as well as present an illustrative application mapping detailed 3-D brain changes in sequential magnetic resonance imaging scans of a patient diagnosed with semantic dementia. Using permutation tests, we show that the symmetrization of image registration statistically reduces skewness in the log-Jacobian map.

Automated ultrasound scanning on a dual-modality breast imaging system: coverage and motion issues and solutions

OBJECTIVE

We are developing an automated ultrasound imaging-mammography system wherein a digital mammography unit has been augmented with a motorized ultrasound transducer carriage above a special compression paddle. Challenges of this system are acquiring complete coverage of the breast and minimizing motion. We assessed these problems and investigated methods to increase coverage and stabilize the compressed breast.

METHODS

Visual tracings of the breast-to-paddle contact area and breast periphery were made for 10 patients to estimate coverage area. Various motion artifacts were evaluated in 6 patients. Nine materials were tested for coupling the paddle to the breast. Fourteen substances were tested for coupling the transducer to the paddle in lateral-to-medial and medial-to-lateral views and filling the gap between the peripheral breast and paddle. In-house image registration software was used to register adjacent ultrasound sweeps.

RESULTS

The average breast contact area was 56%. The average percentage of the peripheral air gap filled with ultrasound gel was 61%. Shallow patient breathing proved equivalent to breath holding, whereas speech and sudden breathing caused unacceptable artifacts. An adhesive spray that preserves image quality was found to be best for coupling the breast to the paddle and minimizing motion. A highly viscous ultrasound gel proved most effective for coupling the transducer to the paddle for lateral-to-medial and medial-to-lateral views and for edge fill-in.

CONCLUSIONS

The challenges of automated ultrasound scanning in a multimodality breast imaging system have been addressed by developing methods to fill in peripheral gaps, minimize patient motion, and register and reconstruct multisweep ultrasound image volumes.

A pattern-theoretic characterization of biological growth

Mathematical and statistical modeling of biological growth is an important problem in medical diagnostics. Here, we seek tools to analyze changes in anatomical parts using images collected over time. We introduce a structured model, called Growth by Random Iterated Diffeomorphisms (GRID), that treats a cumulative growth deformation as a composition of several elementary deformations. Each elementary deformation applies to a small region by capturing deformation local to that region and is characterized by a seed and a radial deformation pattern around that seed. These GRID variables--seed locations and radial deformation patterns---are estimated from observed images in two steps: 1) estimate a cumulative deformation over an observation interval; 2) estimate GRID variables using maximum-likelihood criterion from this estimated cumulative deformation. We demonstrate this framework using an MRI image data of a rat's brain growth. For future statistical analysis, we propose a time-varying Poisson process for the seed placements and a random drawing from a predetermined catalog of deformations for the radial deformation patterns.

3D pattern of brain atrophy in HIV/AIDS visualized using tensor-based morphometry

35% of HIV-infected patients have cognitive impairment, but the profile of HIV-induced brain damage is still not well understood. Here we used tensor-based morphometry (TBM) to visualize brain deficits and clinical/anatomical correlations in HIV/AIDS. To perform TBM, we developed a new MRI-based analysis technique that uses fluid image warping, and a new alpha-entropy-based information-theoretic measure of image correspondence, called the Jensen-Rényi divergence (JRD).

METHODS

3D T1-weighted brain MRIs of 26 AIDS patients (CDC stage C and/or 3 without HIV-associated dementia; 47.2+/-9.8 years; 25M/1F; CD4+ T-cell count: 299.5+/-175.7/microl; log10 plasma viral load: 2.57+/- 1.28 RNA copies/ml) and 14 HIV-seronegative controls (37.6+/-12.2 years; 8M/6F) were fluidly registered by applying forces throughout each deforming image to maximize the JRD between it and a target image (from a control subject). The 3D fluid registration was regularized using the linearized Cauchy-Navier operator. Fine-scale volumetric differences between diagnostic groups were mapped. Regions were identified where brain atrophy correlated with clinical measures.

RESULTS

Severe atrophy ( approximately 15-20% deficit) was detected bilaterally in the primary and association sensorimotor areas. Atrophy of these regions, particularly in the white matter, correlated with cognitive impairment (P = 0.033) and CD4+ T-lymphocyte depletion (P = 0.005).

CONCLUSION

TBM facilitates 3D visualization of AIDS neuropathology in living patients scanned with MRI. Severe atrophy in frontoparietal and striatal areas may underlie early cognitive dysfunction in AIDS patients, and may signal the imminent onset of AIDS dementia complex.

Automated surface matching using mutual information applied to Riemann surface structures

Many medical imaging applications require the computation of dense correspondence vector fields that match one surface with another. To avoid the need for a large set of manually-defined landmarks to constrain these surface correspondences, we developed an algorithm to automate the matching of surface features. It extends the mutual information method to automatically match general 3D surfaces (including surfaces with a branching topology). First, we use holomorphic 1-forms to induce consistent conformal grids on both surfaces. High genus surfaces are mapped to a set of rectangles in the Euclidean plane, and closed genus-zero surfaces are mapped to the sphere. Mutual information is used as a cost functional to drive a fluid flow in the parameter domain that optimally aligns stable geometric features (mean curvature and the conformal factor) in the 2D parameter domains. A diffeomorphic surface-to-surface mapping is then recovered that matches anatomy in 3D. We also present a spectral method that ensures that the grids induced on the target surface remain conformal when pulled through the correspondence field. Using the chain rule, we express the gradient of the mutual information between surfaces in the conformal basis of the source surface. This finite-dimensional linear space generates all conformal reparameterizations of the surface. We apply the method to hippocampal surface registration, a key step in subcortical shape analysis in Alzheimer's disease and schizophrenia.