Papers by Francoise PEYRIN
Scientific Reports, Jun 1, 2017
The weightless environment during spaceflight induces site-specific bone loss. The 30-day Bion-M1... more The weightless environment during spaceflight induces site-specific bone loss. The 30-day Bion-M1 mission offered a unique opportunity to characterize the skeletal changes after spaceflight and an 8-day recovery period in mature male C57/BL6 mice. In the femur metaphysis, spaceflight decreased the trabecular bone volume (−64% vs. Habitat Control), dramatically increased the bone resorption (+140% vs. Habitat Control) and induced marrow adiposity invasion. At the diaphysis, cortical thinning associated with periosteal resorption was observed. In the Flight animal group, the osteocyte lacunae displayed a reduced volume and a more spherical shape (synchrotron radiation analyses), and empty lacunae were highly increased (+344% vs. Habitat Control). Tissue-level mechanical cortical properties (i.e., hardness and modulus) were locally decreased by spaceflight, whereas the mineral characteristics and collagen maturity were unaffected. In the vertebrae, spaceflight decreased the overall bone volume and altered the modulus in the periphery of the trabecular struts. Despite normalized osteoclastic activity and an increased osteoblast number, bone recovery was not observed 8 days after landing. In conclusion, spaceflight induces osteocyte death, which may trigger bone resorption and result in bone mass and microstructural deterioration. Moreover, osteocyte cell death, lacunae mineralization and fatty marrow, which are hallmarks of ageing, may impede tissue maintenance and repair. Spaceflight sojourns induce a series of physiological adaptations in the human body that predispose astronauts to an increased risk of bone fracture after returning to Earth. Under the weightless spaceflight environment, astronauts are subject to factors such as radiation exposure, biorhythm changes, and vestibular alterations, and spaceflight induces site-specific bone loss in both humans and rodents 1-5. Previous in-flight animal research conducted on young growing rats during relatively short-term missions (from 4 to 19 days) has considerably advanced our understanding of skeletal adaptations to spaceflight, including the mechanisms of bone loss and skeletal development in microgravity 6. These experiments showed either deterioration in trabecular and cortical bone parameters 7-11 or no changes 12-16 depending on the animal's age, strain levels, habitat, flight duration and the delay between landing and sample collection. Nevertheless, in the context
2020 28th European Signal Processing Conference (EUSIPCO)
Denoising algorithms via sparse representation are among the state-of-the art for image restorati... more Denoising algorithms via sparse representation are among the state-of-the art for image restoration. On previous work, we proposed SPADE-a sparse-and prior-based method for 3D-image denoising. In this work, we extend this idea to learning approaches and propose a novel residual-U-Net prior-based (ResPrU-Net) method that exploits a prior image. The proposed ResPrU-Net architecture has two inputs, the noisy image and the prior image, and a residual connection that connects the prior image to the output of the network. We compare ResPrU-Net to U-Net and SPADE on human knee data acquired on a spectral computerized tomography scanner. The prior image is built from the noisy image by combining information from neighbor slices and it is the same for both SPADE and ResPrU-Net. For deep learning approaches, we use four knee samples and data augmentation for training, one knee for validation and two for test. Results show that for high noise, U-Net leads to worst results, with images that are excessively blurred. Priorbased methods, SPADE and ResPrU-Net, outperformed U-Net, leading to restored images that present similar image quality than the target. ResPrU-Net provides slightly better results than SPADE. For low noise, methods present similar results.
HAL (Le Centre pour la Communication Scientifique Directe), Jul 7, 2019
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI), 2020
We propose an algorithm based on marker-controlled watershed and the monogenic signal phase asymm... more We propose an algorithm based on marker-controlled watershed and the monogenic signal phase asymmetry for the segmentation of bone and micro-vessels in mouse bone. The images are acquired using synchrotron radiation micro-computed tomography (SR-µCT). The marker image is generated with hysteresis thresholding and morphological filters. The control surface is generated using the phase asymmetry of the monogenic signal in order to detect edgelike structures only, as well as improving detection in low contrast areas, such as bone-vessel interfaces. The quality of segmentation is evaluated by comparing to manually segmented images using the Dice coefficient. The proposed method shows substantial improvement compared to a previously proposed method based on hysteresis thresholding, as well as compared to using the gradient image as control surface. The algorithm was applied to images of healthy and metastatic bone, permitting quantification of both bone and vessel structures.
The Insight Journal, 2011
This document illustrates how to process large images (5 and 23 Gigabytes in size) by taking adva... more This document illustrates how to process large images (5 and 23 Gigabytes in size) by taking advantage of the streaming capabilities of the Insight Toolkit ITK. Here we illustrate two scenarios: (a) the case when the image itself is larger than the computer’s RAM, (b) the case when the image is large but still can fit in the computer’s RAM. This report is intended to serve as a tutorial on how to take advantage of this memory management capabilities of ITK version 4.This paper is accompanied with the source code, input data, parameters and output data that we used for validating the algorithm described in this paper. This adheres to the fundamental principle that scientific publications must facilitate reproducibility of the reported results.
Medical Physics, 2022
Purpose: Computed tomography (CT) is a technique of choice to image bone structure at different s... more Purpose: Computed tomography (CT) is a technique of choice to image bone structure at different scales. Methods to enhance the quality of degraded reconstructions obtained from low-dose CT data have shown impressive results recently, especially in the realm of supervised deep learning. As the choice of the loss function affects the reconstruction quality, it is necessary to focus on the way neural networks evaluate the correspondence between predicted and target images during the training stage. This is even more true in the case of bone microarchitecture imaging at high spatial resolution where both the quantitative analysis of Bone Mineral Density (BMD) and bone microstructure are essential for assessing diseases such as osteoporosis. Our aim is thus to evaluate the quality of reconstruction on key metrics for diagnosis depending on the loss function that has been used for training the neural network. Methods: We compare and analyze volumes that are reconstructed with neural networks trained with pixelwise, structural and adversarial loss functions or with a combination of them. We perform realistic simulations of various low-dose acquisitions of bone microarchitecture. Our comparative study is performed with metrics that have an interest regarding the diagnosis of bone diseases. We therefore focus on bone-specific metrics such as BV/TV, resolution, connectivity assessed with the Euler number and quantitative analysis of BMD to evaluate the quality of reconstruction obtained with networks trained with the different loss functions. Results: We find that using L1 norm as the pixelwise loss is the best choice compared to L2 or no pixelwise loss since it improves resolution without deteriorating other metrics. VGG perceptual loss, especially when combined with an adversarial loss, allows to better retrieve topological and morphological parameters of bone microarchitecture compared to SSIM. This however leads to a decreased resolution performance. The adversarial loss enchances the reconstruction performance in terms of BMD distribution accuracy. Conclusions: In order to retrieve the quantitative and structural characteristics of bone microarchitecture that are essential for post-reconstruction diagnosis, our results suggest to use L1 norm as part of the loss function. Then, trade-offs should be made depending on the application: VGG perceptual loss improves accuracy in terms of connectivity at the cost of a deteriorated resolution, and adversarial losses help better retrieve BMD distribution while significantly increasing the training time.
European Radiology Experimental, 2022
Background Dual-energy computed tomography has shown a great interest for musculoskeletal patholo... more Background Dual-energy computed tomography has shown a great interest for musculoskeletal pathologies. Photon-counting spectral computed tomography (PCSCT) can acquire data in multiple energy bins with the potential to increase contrast, especially for soft tissues. Our objectives were to assess the value of PCSST to characterise cartilage and to extract quantitative measures of subchondral bone integrity. Methods Seven excised human knees (3 males and 4 females; 4 normal and 3 with osteoarthritis; age 80.6 ± 14 years, mean ± standard deviation) were scanned using a clinical PCSCT prototype scanner. Tomographic image reconstruction was performed after Compton/photoelectric decomposition. Virtual monoenergetic images were generated from 40 keV to 110 keV every 10 keV (cubic voxel size 250 × 250 × 250 μm3). After selecting an optimal virtual monoenergetic image, we analysed the grey level histograms of different tissues and extracted quantitative measurements on bone cysts. Results Th...
2019 27th European Signal Processing Conference (EUSIPCO), 2019
Denoising algorithms via sparse representation are among the state-of-the art for 2D image restor... more Denoising algorithms via sparse representation are among the state-of-the art for 2D image restoration. In this work, we propose a novel sparse and prior-based algorithm for 3D image denoising (SPADE). SPADE is a modification of total variation (TV) problem with an additional functional that promotes sparsity with respect to a prior image. The prior is obtained from the noisy image by combining information from neighbor slices. The functional is minimized using the split Bregman method, which leads to an efficient method for large scale 3D denoising, with computational cost given by three FFT per iteration. SPADE is compared to TV and dictionary learning on the Shepp-Logan phantom and on human knee data acquired on a spectral computerized tomography scanner. SPADE converges in approximately ten iterations and provides comparable or better results than the other methods. In addition, the exploitation of the prior image avoids the patchy, cartoon-like images provided by TV and provides a more natural texture.
Skeletal diseases such as osteoporosis constitute a seri-ous socio-economic burden in aging socie... more Skeletal diseases such as osteoporosis constitute a seri-ous socio-economic burden in aging societies. Patient specific tissue strength estimation could be helpful for personalised treatment strategies. Reference point in-dentation (RPI) [1] that derives from instrumented in-dentation [2] has been proposed. With imprints of sev-eral 100 μm at indentation depths of ~80 μm several structural units are affected. It is necessary to better un-derstand the processes under the indenter tip but studies investigating the impact of larger indenters are currently missing. Combining high resolution imaging with in situ microindentation could help to investigate the zone un-derneath the indenter and observe emerging cracks. We aimed at (i) combining microindentations with synchro-tron radiation micro-computed tomography (SRμCT) to investigate (ii) crack systems under two different indent-ers (iii) in axial or transverse testing direction.
The osteocyte network in bone has attracted great interest due to the role of osteocytes in mecha... more The osteocyte network in bone has attracted great interest due to the role of osteocytes in mechanosensing and regulation of bone remodeling. Osteocytes reside in lacunae and are interconnected by cellular processes running through a network of canaliculi; canals roughly 200 nm in diameter. The canalicular network plays a vital role in the communication between osteocytes and facilitates a way for osteocytes to orchestrate bone remodelling. Rodents are widely used as model organisms to study experimentally induced effects in bone. Human and rodent bone does, however, display large structural variations with the largest difference being the absence of harversian remodeling in rodents, which has profound implications for bone microstructure [1]. Here we have studied the lacuna-canalicular network in mouse bone to describe the communication network and the structural features found on the sub-micro meter length scale. Describing the hierarchical structure of bone demands multiscale ima...
International Journal of Tomography and Simulation, 2016
In this work, we address the problem of the reconstruction of binary images from a small number o... more In this work, we address the problem of the reconstruction of binary images from a small number of noisy tomographic projections. Recently, a new stochastic level-set approach was investigated to refine the reconstruction. The main limitation of this method is that it is only changing the boundaries of the reconstructed regions. In this work, we study a new stochastic approach based on Total Variation (TV) regularization with box constraints. The main advantage of this method is that random shape and boundaries variations can be included in a new way and that topology changes can be also added. The methods are tested on two complex bone micro-CT cross-section images for different noise levels and number of projections. While for the higher noise levels, the best reconstructions are obtained with a stochastic diffusion based on the Total Variation regularization, large decreases of the reconstruction errors are obtained when shape and topology noises are used simultaneously.
Most of the fractures occur at the radius or at the femoral neck after a fall, involving high str... more Most of the fractures occur at the radius or at the femoral neck after a fall, involving high strain rate (10-1.s-1) [1]. Cortical bone toughness is usually measured under quasi-static loading condition even though it is known to have viscoelastic properties [2]. A recent study on one subject actually showed that the toughness of femoral shaft specimens is lower at a strain rate representative of a fall [3]. We hypothesize this effect could be the same whatever the anatomical location. The aim of this study is thus to quantify human cortical bone toughness using paired radii and femurs (neck and shaft) considering two strain rates (quasi-static and representative of a fall).
Biomedical Optics Express, 2021
While numerous transgenic mouse strains have been produced to model the formation of amyloid-β (A... more While numerous transgenic mouse strains have been produced to model the formation of amyloid-β (Aβ) plaques in the brain, efficient methods for whole-brain 3D analysis of Aβ deposits have to be validated and standardized. Moreover, routine immunohistochemistry performed on brain slices precludes any shape analysis of Aβ plaques, or require complex procedures for serial acquisition and reconstruction. The present study shows how in-line (propagation-based) X-ray phase-contrast tomography (XPCT) combined with ethanol-induced brain sample dehydration enables hippocampus-wide detection and morphometric analysis of Aβ plaques. Performed in three distinct Alzheimer mouse strains, the proposed workflow identified differences in signal intensity and 3D shape parameters: 3xTg displayed a different type of Aβ plaques, with a larger volume and area, greater elongation, flatness and mean breadth, and more intense average signal than J20 and APP/PS1. As a label-free non-destructive technique, XP...
IEEE Access, 2021
The state-of-the art for solving the nonlinear material decomposition problem in spectral compute... more The state-of-the art for solving the nonlinear material decomposition problem in spectral computed tomography is based on variational methods, but these are computationally slow and critically depend on the particular choice of the regularization functional. Convolutional neural networks have been proposed for addressing these issues. However, learning algorithms require large amounts of experimental data sets. We propose a deep learning strategy for solving the material decomposition problem based on a U-Net architecture and a Sim2Real transfer learning approach where the knowledge that we learn from synthetic data is transferred to a real-world scenario. In order for this approach to work, synthetic data must be realistic and representative of the experimental data. For this purpose, numerical phantoms are generated from human CT volumes of the KiTS19 Challenge dataset, segmented into specific materials (soft tissue and bone). These volumes are projected into sinogram space in order to simulate photon counting data, taking into account the energy response of the scanner. We compared projection-and image-based decomposition approaches where the network is trained to decompose the materials either in the projection or in the image domain. The proposed Sim2Real transfer strategies are compared to a regularized Gauss-Newton (RGN) method on synthetic data, experimental phantom data and human thorax data. INDEX TERMS Spectral CT, inverse problem, deep learning, transfer learning.
Journal of The Royal Society Interface, 2019
With ageing and various diseases, the vascular pore volume fraction (porosity) in cortical bone i... more With ageing and various diseases, the vascular pore volume fraction (porosity) in cortical bone increases, and the morphology of the pore network is altered. Cortical bone elasticity is known to decrease with increasing porosity, but the effect of the microstructure is largely unknown, while it has been thoroughly studied for trabecular bone. Also, popular micromechanical models have disregarded several micro-architectural features, idealizing pores as cylinders aligned with the axis of the diaphysis. The aim of this paper is to quantify the relative effects on cortical bone anisotropic elasticity of porosity and other descriptors of the pore network micro-architecture associated with pore number, size and shape. The five stiffness constants of bone assumed to be a transversely isotropic material were measured with resonant ultrasound spectroscopy in 55 specimens from the femoral diaphysis of 29 donors. The pore network, imaged with synchrotron radiation X-ray micro-computed tomogra...
2015 23rd European Signal Processing Conference (EUSIPCO), 2015
The investigation of bone fragility diseases, as osteoporosis, is based on the analysis of the tr... more The investigation of bone fragility diseases, as osteoporosis, is based on the analysis of the trabecular bone microarchitecture. The aim of this paper is to improve the in-vivo trabecular bone segmentation and quantification by increasing the resolution of bone micro-architecture images. We propose a semi-blind joint super-resolution/segmentation approach based on a Total Variation regularization with a convex constraint. A comparison with the bicubic interpolation method and the non-blind version of the proposed method is shown. The validation is performed on blurred, noisy and down-sampled 3D synchrotron micro-CT bone images. Good estimates of the blur and of the high resolution image are obtained with the semi-blind approach. Preliminary results are obtained with the semi-blind approach on real HR-pQCT images.
Congrès sous l’égide de la Société Française de Génie Biologique et Médical (SFGBM).
-L'objectif de ce travail est de proposer une méthode de quantification des structures poreuses d... more -L'objectif de ce travail est de proposer une méthode de quantification des structures poreuses de mousses solides à partir d'images tomographiques à très haute résolution. Nous avons imagé des mousses polymères directement en 3D grâce à une technique de microtomographie par rayonnement synchrotron avec une résolution spatiale de 6.65 microns. Après la phase d'acquisition, des paramètres quantitatifs tels que l'épaisseur des parois de la matrices et le diamètre des bulles d'air doivent être mesurés. Nous comparons pour cela trois approches différentes : par ouverture, par épaisseur locale et par la méthode des sécantes (MIL). Ceci permet de valider les résultats des différentes méthodes. Cette quantification a été appliquée à la mesure sur image 3D des variations dues au changement de température dans le procédé de fabrication des mousses polymères.
Elasticity and porosity in human cortical bone: models and experiments
Uploads
Papers by Francoise PEYRIN