Thierry Cresson

Method for fast and accurate segmentation processing from prior shape: application to femoral head segmentation on x-ray images

This paper proposes a prior shape segmentation method to create a constant-width ribbon-like zone that runs along the boundary to be extracted. The image data corresponding to that zone is transformed into a rectangular image subspace where the boundary is roughly straightened. Every step of the segmentation process is then applied to that straightened subspace image where the final extracted boundary is transformed back into the original image space. This approach has the advantage of producing very efficient filtering and edge detection using conventional techniques. The final boundary is continuous even over image regions where partial information is missing. The technique was applied to the femoral head segmentation where we show that the final segmented boundary is very similar to the one obtained manually by a trained orthopedist and has low sensitivity to the initial positioning of the prior shape.

Surface reconstruction from planar x-ray images using moving least squares

Planar radiographs still are the gold standard for the measurement of the skeletal weight-bearing shape and posture. In this paper, we propose to use an as-rigid-as-possible deformation approach based on moving least squares to obtain 3D personalized bone models from planar x-ray images. Our prototype implementation is capable of performing interactive rate shape editing. The biplane reconstructions of both femur and vertebrae show a good accuracy when compared to CT-scan.

Performing Accurate Joint Kinematics From 3-D In Vivo Image Sequences Through Consensus-Driven Simultaneous Registration

This paper addresses the problem of the robust registration of multiple observations of the same object. Such a problem typically arises whenever it becomes necessary to recover the trajectory of an evolving object observed through standard 3-D medical imaging techniques. The instances of the tracked object are assumed to be variously truncated, locally subject to morphological evolutions throughout the sequence, and imprinted with significant segmentation errors as well as significant noise perturbations. The algorithm operates through the robust and simultaneous registration of all surface instances of a given object through median consensus. This operation consists of two interwoven processes set up to work in close collaboration. The first one progressively generates a median and implicit shape computed with respect to current estimations of the registration transformations, while the other refines these transformations with respect to the current estimation of their median shape. When compared with standard robust techniques, tests reveal significant improvements, both in robustness and precision. The algorithm is based on widely-used techniques, and proves highly effective while offering great flexibility of utilization.

Coupling 2D/3D registration method and statistical model to perform 3D reconstruction from partial x-rays images data

3D reconstructions of the spine from a frontal and sagittal radiographs is extremely challenging. The overlying features of soft tissues and air cavities interfere with image processing. It is also difficult to obtain information that is accurate enough to reconstruct complete 3D models. To overcome these problems, the proposed method efficiently combines the partial information contained in two images from a patient with a statistical 3D spine model generated from a database of scoliotic patients. The algorithm operates through two simultaneous iterating processes. The first one generates a personalized vertebra model using a 2D/3D registration process with bone boundaries extracted from radiographs, while the other one infers the position and the shape of other vertebrae from the current estimation of the registration process using a statistical 3D model. Experimental evaluations have shown good performances of the proposed approach in terms of accuracy and robustness when compared to CT-scan.

Automatic spine and pelvis detection in frontal X-rays using deep neural networks for patch displacement learning

This paper proposes a method to automatically detect the spine and pelvis structures from a postero-anterior radiograph. From a training dataset, a non-linear regression model was trained using a deep neural network (DNN) in order to predict the displacement that recovers the optimal location of an anatomical landmark from an input image patch. Using a DNN for each landmark of a 2D simplified model of the spine, a detection sequence was able to localize the vertebral body centers and femoral heads. The whole process is regularized using a statistical shape model of a simplified model of the spine. The quantitative assessment on a set of 121 radiographs of scoliotic patients presented a mean localization errors of 3.5 ± 3.6 mm and 5.7 ± 6 mm respectively for the femoral heads and the vertebral body centers (vertebral levels T1 to L5). The mean error for the spinal curve automatic detection was 2 ± 2.8 mm, which is accurate enough to determine a first estimate of the spine 3D reconstruction in a 3D biplanar reconstruction scheme.

3D shape reconstruction of bone from two x-ray images using 2D/3D non-rigid registration based on moving least-squares deformation

Several studies based on biplanar radiography technologies are foreseen as great systems for 3D-reconstruction applications for medical diagnoses. This paper proposes a non-rigid registration method to estimate a 3D personalized shape of bone models from two planar x-ray images using an as-rigid-as-possible deformation approach based on a moving least-squares optimization method. Based on interactive deformation methods, the proposed technique has the ability to let a user improve readily and with simplicity a 3D reconstruction which is an important step in clinical applications. Experimental evaluations of six anatomical femur specimens demonstrate good performances of the proposed approach in terms of accuracy and robustness when compared to CT-scan.

Liver Segmentation on CT and MR Using Laplacian Mesh Optimization

Objective: The purpose of this paper is to describe a semiautomated segmentation method for the liver and evaluate its performance on CT-scan and MR images. Methods: First, an approximate 3-D model of the liver is initialized from a few user-generated contours to globally outline the liver shape. The model is then automatically deformed by a Laplacian mesh optimization scheme until it precisely delineates the patients liver. A correction tool was implemented to allow the user to improve the segmentation until satisfaction. Results: The proposed method was tested against 30 CT-scans from the SLIVER07 challenge repository and 20 MR studies from the Montreal University Hospital Center, covering a wide spectrum of liver morphologies and pathologies. The average volumetric overlap error was 5.1% for CT and 7.6% for MRI and the average segmentation time was 6 min. Conclusion: The obtained results show that the proposed method is efficient, reliable, and could effectively be used routinely in the clinical setting. Significance: The proposed approach can alleviate the cumbersome and tedious process of slice-wise segmentation required for precise hepatic volumetry, virtual surgery, and treatment planning.

Investigation of 3D glenohumeral displacements from 3D reconstruction using biplane X-ray images: Accuracy and reproducibility of the technique and preliminary analysis in rotator cuff tear patients

Rotator cuff (RC) tears may be associated with increased glenohumeral instability; however, this instability is difficult to quantify using currently available diagnostic tools. Recently, the three-dimensional (3D) reconstruction and registration method of the scapula and humeral head, based on sequences of low-dose biplane X-ray images, has been proposed for glenohumeral displacement assessment. This research aimed to evaluate the accuracy and reproducibility of this technique and to investigate its potential with a preliminary application comparing RC tear patients and asymptomatic volunteers. Accuracy was assessed using CT scan model registration on biplane X-ray images for five cadaveric shoulder specimens and showed differences ranging from 0.6 to 1.4mm depending on the direction of interest. Intra- and interobserver reproducibility was assessed through two operators who repeated the reconstruction of five subjects three times, allowing defining 95% confidence interval ranging from ±1.8 to ±3.6mm. Intraclass correlation coefficient varied between 0.84 and 0.98. Comparison between RC tear patients and asymptomatic volunteers showed differences of glenohumeral displacements, especially in the superoinferior direction when shoulder was abducted at 20° and 45°. This study thus assessed the accuracy of the low-dose 3D biplane X-ray reconstruction technique for glenohumeral displacement assessment and showed potential in biomechanical and clinical research.

A method to study 3D knee pseudo-kinematics using low-dose stereoradiography during static squat.

To investigate clinical questions as kinematic or postural differences between healthy and pathological populations, the current trend is to couple imagery to movement or to posture. To do so, researchers generally register a threedimensional (3D) model acquired from either CT scan or magnetic resonance imaging on 2D fluoroscopic images (registration on dynamic activities) or several radiographs (registration on posture or pseudo-kinematics) (Moro-oka et al. 2007). The literature suggests the use of so-called ‘extrinsic 2D/3D registration’. This implies the need of geometrical objects in the radiographic scene. Embedded tantalum beads or prostheses (Scarvell et al. 2010; Sharma et al. 2012),well detectable on the radiographs, are generally used. They allow attaining the expected measurement accuracy ,18 and ,1mm. When no tantalum beads or prostheses are present, the registration process lacks reliability and accuracy because of the difficulty to extract precise information from the radiographic images (blurry contours, superimposition of structures and cylindrical nature of long bones) (Belvedere et al. 2013). The goal of this work was to present an intrinsic 2D/3D registration method and to validate its reliability on patients’ images aswell as its accuracy on simulated radiographs in a pseudo-kinematic context.

Fast 3D reconstruction of the spine from biplanar radiography: a diagnosis tool for routine scoliosis diagnosis and research in biomechanics

Reconstruction methods from biplanar radiography allow a 3D clinical analysis, for patients in standing position, with a low radiation dose. This low-dose and postural imaging modality is thus very interesting for scoliosis clinical diagnosis and research in biomechanics. Nevertheless, such applications require both accurate and fast 3D reconstruction methods. Fast approaches, based on statistical models (Pomero et al. 2004; Gille et al. 2007), allow to obtain an estimate of the spinal 3D reconstruction within 14min. However, this remains too tedious to be used in a clinical routine. The purpose of this study is to propose and evaluate a novel semi-automated 3D reconstruction method of the spine from biplanar radiography. This method relies on a parametric model of the spine using statistical inferences and automatic registration methods based on image processing. Two reconstruction levels are proposed: a first reconstruction level (‘Fast Spine’), providing a fast estimate of the 3D reconstruction and accurate clinical measurements, dedicated to a routine clinical use, and a more accurate second reconstruction level (‘Full Spine’) for applications in biomechanical research.