Search

Abstract Corneal guttae, which are the abnormal growth of extracellular matrix in the corneal endothelium, are observed in specular images as black droplets that occlude the endothelial cells. To estimate the corneal parameters (endothelial cell density [ECD], coefficient of variation [CV], and hexagonality [HEX]), we propose a new deep learning method that includes a novel attention mechanism (named fNLA), which helps to infer the cell edges in the occluded areas. The approach first derives the cell edges, then infers the well-detected cells, and finally employs a postprocessing method to fix mistakes. This results in a binary segmentation from which the corneal parameters are estimated. We analyzed 1203 images (500 contained guttae) obtained with a Topcon SP-1P microscope. To generate the ground truth, we performed manual segmentation in all images. Several networks were evaluated (UNet, ResUNeXt, DenseUNets, UNet++, etc.) and we found that DenseUNets with fNLA provided the lowest error: a mean absolute error of 23.16 [cells/mm $$^{2}$$ 2 ] in ECD, 1.28 [%] in CV, and 3.13 [%] in HEX. Compared with Topcon’s built-in software, our error was 3–6 times smaller. Overall, our approach handled notably well the cells affected by guttae, detecting cell edges partially occluded by small guttae and discarding large areas covered by extensive guttae.

Abstract Background Corneal endothelium (CE) images provide valuable clinical information regarding the health state of the cornea. Computation of the clinical morphometric parameters requires the segmentation of endothelial cell images. Current techniques to image the endothelium in vivo deliver low quality images, which makes automatic segmentation a complicated task. Here, we present two convolutional neural networks (CNN) to segment CE images: a global fully convolutional approach based on U-net, and a local sliding-window network (SW-net). We propose to use probabilistic labels instead of binary, we evaluate a preprocessing method to enhance the contrast of images, and we introduce a postprocessing method based on Fourier analysis and watershed to convert the CNN output images into the final cell segmentation. Both methods are applied to 50 images acquired with an SP-1P Topcon specular microscope. Estimates are compared against a manual delineation made by a trained observer. Results U-net (AUC=0.9938) yields slightly sharper, clearer images than SW-net (AUC=0.9921). After postprocessing, U-net obtains a DICE=0.981 and a MHD=0.22 (modified Hausdorff distance), whereas SW-net yields a DICE=0.978 and a MHD=0.30. U-net generates a wrong cell segmentation in only 0.48% of the cells, versus 0.92% for the SW-net. U-net achieves statistically significant better precision and accuracy than both, Topcon and SW-net, for the estimates of three clinical parameters: cell density (ECD), polymegethism (CV), and pleomorphism (HEX). The mean relative error in U-net for the parameters is 0.4% in ECD, 2.8% in CV, and 1.3% in HEX. The computation time to segment an image and estimate the parameters is barely a few seconds. Conclusions Both methods presented here provide a statistically significant improvement over the state of the art. U-net has reached the smallest error rate. We suggest a segmentation refinement based on our previous work to further improve the performance.

Better insight into white matter (WM) alterations after stroke onset could help to understand the underlying recovery mechanisms and improve future interventions. MR diffusion imaging enables to assess such changes. Our goal was to investigate the relation of WM diffusion characteristics derived from diffusion models of increasing complexity with the motor function of the upper limb. Moreover, we aimed to evaluate the variation of such characteristics across different WM structures of chronic stroke patients in comparison to healthy subjects. Subjects were scanned with a two b-value diffusion-weighted MRI protocol to exploit multiple diffusion models: single tensor, single tensor with isotropic compartment, bi-tensor model, bi-tensor with isotropic compartment. From each model we derived the mean tract fractional anisotropy (FA), mean (MD), radial (RD) and axial (AD) diffusivities outside the lesion site based on a WM tracts atlas. Asymmetry of these measures was correlated with the Fugl-Meyer upper extremity assessment (FMA) score and compared between patient and control groups. Eighteen chronic stroke patients and eight age-matched healthy individuals participated in the study. Significant correlation of the outcome measures with the clinical scores of stroke recovery was found. The lowest correlation of the corticospinal tract FAasymmetry and FMA was with the single tensor model (r = −0.3, p = 0.2) whereas the other models reported results in the range of r = −0.79 ÷ −0.81 and p = 4E-5 ÷ 8E-5. The corticospinal tract and superior longitudinal fasciculus showed most alterations in our patient group relative to controls. Multiple compartment models yielded superior correlation of the diffusion measures and FMA compared to the single tensor model.

PurposePharmacokinetic models facilitate assessment of properties of the micro-vascularization based on DCE-MRI data. However, accurate pharmacokinetic modeling in the liver is challenging since it has two vascular inputs and it is subject to large deformation and displacement due to respiration.MethodsWe propose an improved pharmacokinetic model for the liver that (1) analytically models the arrival-time of the contrast agent for both inputs separately; (2) implicitly compensates for signal fluctuations that can be modeled by varying applied flip-angle e.g. due to B1-inhomogeneity. Orton's AIF model is used to analytically represent the vascular input functions. The inputs are independently embedded into the Sourbron model. B1-inhomogeneity-driven variations of flip-angles are accounted for to justify the voxel's displacement with respect to a pre-contrast image.ResultsThe new model was shown to yield lower root mean square error (RMSE) after fitting the model to all but a minority of voxels compared to Sourbron's approach. Furthermore, it outperformed this existing model in the majority of voxels according to three model-selection criteria.ConclusionOur work primarily targeted to improve pharmacokinetic modeling for DCE-MRI of the liver. However, other types of pharmacokinetic models may also benefit from our approaches, since the techniques are generally applicable.

Efficient design of multi-component food products containing dry and wet components such as biscuits with a moist fruit filling, is of growing interests for food industry. Technology is needed to prevent or reduce water migration from the moist filling to the dry porous cereal material. This can be done by using moisture barrier systems. Knowledge of the microstructure and its relation to water mobility is necessary to develop stable products. This paper describes a study that uses X-ray microtomography (μCT) for the characterisation and visualisation of the 3-D structure of crackers with different porosity, coated biscuit shells and soup inclusions. μCT was used for imaging the inner cellular structure of the cereal matrix or to analyse the integrity of moisture barriers applied on the cereal product. 3-D image analysis methods were developed to obtain quantitative information about the cellular matrix which can be used as input for simulation models for moisture migration. The developed 3-D image analysis method maps the open cellular structure onto a network (graph) representation in which the nodes correspond to the pores and the vertices to the pore-topore interconnection. The pores (nodes) have properties such as volume, surface area and location whereas the vertices have properties such as direct (open connection) and indirect (separated by a single lamella) area. To check the segmentation and network description a model for pore to pore resistance was used. The obtained results demonstrate the potential of μCT and 3-D image analysis for extracting structural information which can be used in models for the moisture penetration in a cellular bakery product.

This paper presents and studies a framework for reliable modeling of diffusion MRI using a data-acquisition adaptive prior.Automated relevance determination estimates the mean of the posterior distribution of a rank-2 dual tensor model exploiting Jeffreys prior (JARD). This data-acquisition prior is based on the Fisher information matrix and enables the assessment whether two tensors are mandatory to describe the data. The method is compared to Maximum Likelihood Estimation (MLE) of the dual tensor model and to FSL's ball-and-stick approach.Monte Carlo experiments demonstrated that JARD's volume fractions correlated well with the ground truth for single and crossing fiber configurations. In single fiber configurations JARD automatically reduced the volume fraction of one compartment to (almost) zero. The variance in fractional anisotropy (FA) of the main tensor component was thereby reduced compared to MLE. JARD and MLE gave a comparable outcome in data simulating crossing fibers. On brain data, JARD yielded a smaller spread in FA along the corpus callosum compared to MLE. Tract-based spatial statistics demonstrated a higher sensitivity in detecting age-related white matter atrophy using JARD compared to both MLE and the ball-and-stick approach.The proposed framework offers accurate and precise estimation of diffusion properties in single and dual fiber regions.