Fully automatic segmentation of AP pelvis X-rays via random forest regression with efficient feature selection and hierarchical sparse shape composition

In clinical practice, traditional X-ray radiography is widely used, and knowledge of landmarks and contours in anteroposterior (AP) pelvis X-rays is invaluable for computer aided diagnosis, hip surgery planning and image-guided interventions. This paper presents a fully automatic approach for landmark detection and shape segmentation of both pelvis and femur in conventional AP X-ray images. Our approach is based on the framework of landmark detection via Random Forest (RF) regression and shape regularization via hierarchical sparse shape composition. We propose a visual feature FL-HoG (Flexible- Level Histogram of Oriented Gradients) and a feature selection algorithm based on trace radio optimization to improve the robustness and the efficacy of RF-based landmark detection. The landmark detection result is then used in a hierarchical sparse shape composition framework for shape regularization. Finally, the extracted shape contour is fine-tuned by a post-processing step based on low level image features. The experimental results demonstrate that our feature selection algorithm reduces the feature dimension in a factor of 40 and improves both training and test efficiency. Further experiments conducted on 436 clinical AP pelvis X-rays show that our approach achieves an average point-to-curve error around 1.2 mm for femur and 1.9 mm for pelvis.

Is the acetabular cup orientation after total hip arthroplasty on a two dimension or three dimension model accurate?

Purpose Malposition of the acetabular component in total hip arthroplasty (THA) is a common surgical problem that can lead to hip dislocation, reduced range of motion and may result in early loosening. The aim of this study is to validate the accuracy and reproducibility of a single x-ray image based 2D/3D reconstruction technique in determining cup inclination and anteversion against two different computer tomography (CT)-based measurement techniques. Methods Cup anteversion and inclination of 20 patients after cementless primary THA was measured on standard anteroposterior (AP) radiographs with the help of the single x-ray 2D/3D reconstruction program and compared with two different 3D CT-based analyses [Ground Truth (GT) and MeVis (MV) reconstruction model]. Results The measurements from the single x-ray 2D/3D reconstruction technique were strongly correlated with both types of CT image-processing protocols for both cup inclination [R²=0.69 (GT); R²=0.59 (MV)] and anteversion [R²=0.89 (GT); R²=0.80 (MV)]. Conclusions The single x-ray image based 2D/3D reconstruction technique is a feasible method to assess cup position on postoperative x-rays. CTscans remain the golden standard for a more complex biomechanical evaluation when a lower tolerance limit (+/-2 degrees) is required.

Development of a balanced experimental–computational approach to understanding the mechanics of proximal femur fractures

The majority of people who sustain hip fractures after a fall to the side would not have been identified using current screening techniques such as areal bone mineral density. Identifying them, however, is essential so that appropriate pharmacological or lifestyle interventions can be implemented. A protocol, demonstrated on a single specimen, is introduced, comprising the following components; in vitro biofidelic drop tower testing of a proximal femur; high-speed image analysis through digital image correlation; detailed accounting of the energy present during the drop tower test; organ level finite element simulations of the drop tower test; micro level finite element simulations of critical volumes of interest in the trabecular bone. Fracture in the femoral specimen initiated in the superior part of the neck. Measured fracture load was 3760 N, compared to 4871 N predicted based on the finite element analysis. Digital image correlation showed compressive surface strains as high as 7.1% prior to fracture. Voxel level results were consistent with high-speed video data and helped identify hidden local structural weaknesses. We found using a drop tower test protocol that a femoral neck fracture can be created with a fall velocity and energy representative of a sideways fall from standing. Additionally, we found that the nested explicit finite element method used allowed us to identify local structural weaknesses associated with femur fracture initiation.

Automatic X-ray landmark detection and shape segmentation via data-driven joint estimation of image displacements

In this paper, we propose a new method for fully-automatic landmark detection and shape segmentation in X-ray images. To detect landmarks, we estimate the displacements from some randomly sampled image patches to the (unknown) landmark positions, and then we integrate these predictions via a voting scheme. Our key contribution is a new algorithm for estimating these displacements. Different from other methods where each image patch independently predicts its displacement, we jointly estimate the displacements from all patches together in a data driven way, by considering not only the training data but also geometric constraints on the test image. The displacements estimation is formulated as a convex optimization problem that can be solved efficiently. Finally, we use the sparse shape composition model as the a priori information to regularize the landmark positions and thus generate the segmented shape contour. We validate our method on X-ray image datasets of three different anatomical structures: complete femur, proximal femur and pelvis. Experiments show that our method is accurate and robust in landmark detection, and, combined with the shape model, gives a better or comparable performance in shape segmentation compared to state-of-the art methods. Finally, a preliminary study using CT data shows the extensibility of our method to 3D data.

FACTS: Fully Automatic CT Segmentation of a Hip Joint

Extraction of surface models of a hip joint from CT data is a pre-requisite step for computer assisted diagnosis and planning (CADP) of periacetabular osteotomy (PAO). Most of existing CADP systems are based on manual segmentation, which is time-consuming and hard to achieve reproducible results. In this paper, we present a Fully Automatic CT Segmentation (FACTS) approach to simultaneously extract both pelvic and femoral models. Our approach works by combining fast random forest (RF) regression based landmark detection, multi-atlas based segmentation, with articulated statistical shape model (aSSM) based fitting. The two fundamental contributions of our approach are: (1) an improved fast Gaussian transform (IFGT) is used within the RF regression framework for a fast and accurate landmark detection, which then allows for a fully automatic initialization of the multi-atlas based segmentation; and (2) aSSM based fitting is used to preserve hip joint structure and to avoid penetration between the pelvic and femoral models. Taking manual segmentation as the ground truth, we evaluated the present approach on 30 hip CT images (60 hips) with a 6-fold cross validation. When the present approach was compared to manual segmentation, a mean segmentation accuracy of 0.40, 0.36, and 0.36 mm was found for the pelvis, the left proximal femur, and the right proximal femur, respectively. When the models derived from both segmentations were used to compute the PAO diagnosis parameters, a difference of 2.0 ± 1.5°, 2.1 ± 1.6°, and 3.5 ± 2.3% were found for anteversion, inclination, and acetabular coverage, respectively. The achieved accuracy is regarded as clinically accurate enough for our target applications.

Localization and Segmentation of 3D Intervertebral Discs in MR Images by Data Driven Estimation

This paper addresses the problem of fully-automatic localization and segmentation of 3D intervertebral discs (IVDs) from MR images. Our method contains two steps, where we first localize the center of each IVD, and then segment IVDs by classifying image pixels around each disc center as foreground (disc) or background. The disc localization is done by estimating the image displacements from a set of randomly sampled 3D image patches to the disc center. The image displacements are estimated by jointly optimizing the training and test displacement values in a data-driven way, where we take into consideration both the training data and the geometric constraint on the test image. After the disc centers are localized, we segment the discs by classifying image pixels around disc centers as background or foreground. The classification is done in a similar data-driven approach as we used for localization, but in this segmentation case we are aiming to estimate the foreground/background probability of each pixel instead of the image displacements. In addition, an extra neighborhood smooth constraint is introduced to enforce the local smoothness of the label field. Our method is validated on 3D T2-weighted turbo spin echo MR images of 35 patients from two different studies. Experiments show that compared to state of the art, our method achieves better or comparable results. Specifically, we achieve for localization a mean error of 1.6-2.0 mm, and for segmentation a mean Dice metric of 85%-88% and a mean surface distance of 1.3-1.4 mm.

Evaluation and Comparison of Anatomical Landmark Detection Methods for Cephalometric X-Ray Images: A Grand Challenge

Cephalometric analysis is an essential clinical and research tool in orthodontics for the orthodontic analysis and treatment planning. This paper presents the evaluation of the methods submitted to the Automatic Cephalometric X-Ray Landmark Detection Challenge, held at the IEEE International Symposium on Biomedical Imaging 2014 with an on-site competition. The challenge was set to explore and compare automatic landmark detection methods in application to cephalometric X-ray images. Methods were evaluated on a common database including cephalograms of 300 patients aged six to 60 years, collected from the Dental Department, Tri-Service General Hospital, Taiwan, and manually marked anatomical landmarks as the ground truth data, generated by two experienced medical doctors. Quantitative evaluation was performed to compare the results of a representative selection of current methods submitted to the challenge. Experimental results show that three methods are able to achieve detection rates greater than 80% using the 4 mm precision range, but only one method achieves a detection rate greater than 70% using the 2 mm precision range, which is the acceptable precision range in clinical practice. The study provides insights into the performance of different landmark detection approaches under real-world conditions and highlights achievements and limitations of current image analysis techniques.

A complete-pelvis segmentation framework for image-free total hip arthroplasty (THA): methodology and clinical study

Background Complete-pelvis segmentation in antero-posterior pelvic radiographs is required to create a patient-specific three-dimensional pelvis model for surgical planning and postoperative assessment in image-free navigation of total hip arthroplasty. Methods A fast and robust framework for accurately segmenting the complete pelvis is presented, consisting of two consecutive modules. In the first module, a three-stage method was developed to delineate the left hemipelvis based on statistical appearance and shape models. To handle complex pelvic structures, anatomy-specific information processing techniques were employed. As the input to the second module, the delineated left hemi-pelvis was then reflected about an estimated symmetry line of the radiograph to initialize the right hemi-pelvis segmentation. The right hemi-pelvis was segmented by the same three-stage method, Results Two experiments conducted on respectively 143 and 40 AP radiographs demonstrated a mean segmentation accuracy of 1.61±0.68 mm. A clinical study to investigate the postoperative assessment of acetabular cup orientations based on the proposed framework revealed an average accuracy of 1.2°±0.9° and 1.6°±1.4° for anteversion and inclination, respectively. Delineation of each radiograph costs less than one minute. Conclusions Despite further validation needed, the preliminary results implied the underlying clinical applicability of the proposed framework for image-free THA.

Ruler Based Automatic C-Arm Image Stitching Without Overlapping Constraint

In this paper, we propose a new method for stitching multiple fluoroscopic images taken by a C-arm instrument. We employ an X-ray radiolucent ruler with numbered graduations while acquiring the images, and the image stitching is based on detecting and matching ruler parts in the images to the corresponding parts of a virtual ruler. To achieve this goal, we first detect the regular spaced graduations on the ruler and the numbers. After graduation labeling, for each image, we have the location and the associated number for every graduation on the ruler. Then, we initialize the panoramic X-ray image with the virtual ruler, and we “paste” each image by aligning the detected ruler part on the original image, to the corresponding part of the virtual ruler on the panoramic image. Our method is based on ruler matching but without the requirement of matching similar feature points in pairwise images, and thus, we do not necessarily require overlap between the images. We tested our method on eight different datasets of X-ray images, including long bones and a complete spine. Qualitative and quantitative experiments show that our method achieves good results.

Unstable argininosuccinate lyase in variant forms of the urea cycle disorder argininosuccinic aciduria.

Loss of function of the urea cycle enzyme argininosuccinate lyase (ASL) is caused by mutations in the ASL gene leading to ASL deficiency (ASLD). ASLD has a broad clinical spectrum ranging from life-threatening severe neonatal to asymptomatic forms. Different levels of residual ASL activity probably contribute to the phenotypic variability but reliable expression systems allowing clinically useful conclusions are not yet available. In order to define the molecular characteristics underlying the phenotypic variability, we investigated all ASL mutations that were hitherto identified in patients with late onset or mild clinical and biochemical courses by ASL expression in human embryonic kidney 293 T cells. We found residual activities >3 % of ASL wild type (WT) in nine of 11 ASL mutations. Six ASL mutations (p.Arg95Cys, p.Ile100Thr, p.Val178Met, p.Glu189Gly, p.Val335Leu, and p.Arg379Cys) with residual activities ≥16 % of ASL WT showed no significant or less than twofold reduced Km values, but displayed thermal instability. Computational structural analysis supported the biochemical findings by revealing multiple effects including protein instability, disruption of ionic interactions and hydrogen bonds between residues in the monomeric form of the protein, and disruption of contacts between adjacent monomeric units in the ASL tetramer. These findings suggest that the clinical and biochemical course in variant forms of ASLD is associated with relevant residual levels of ASL activity as well as instability of mutant ASL proteins. Since about 30 % of known ASLD genotypes are affected by mutations studied here, ASLD should be considered as a candidate for chaperone treatment to improve mutant protein stability.

Computer Vision-Based Carbohydrate Estimation for Type 1 Diabetic Patients Using Smartphones

Background: Individuals with type 1 diabetes (T1D) have to count the carbohydrates (CHOs) of their meal to estimate the prandial insulin dose needed to compensate for the meal’s effect on blood glucose levels. CHO counting is very challenging but also crucial, since an error of 20 grams can substantially impair postprandial control. Method: The GoCARB system is a smartphone application designed to support T1D patients with CHO counting of nonpacked foods. In a typical scenario, the user places a reference card next to the dish and acquires 2 images with his/her smartphone. From these images, the plate is detected and the different food items on the plate are automatically segmented and recognized, while their 3D shape is reconstructed. Finally, the food volumes are calculated and the CHO content is estimated by combining the previous results and using the USDA nutritional database. Results: To evaluate the proposed system, a set of 24 multi-food dishes was used. For each dish, 3 pairs of images were taken and for each pair, the system was applied 4 times. The mean absolute percentage error in CHO estimation was 10 ± 12%, which led to a mean absolute error of 6 ± 8 CHO grams for normal-sized dishes. Conclusion: The laboratory experiments demonstrated the feasibility of the GoCARB prototype system since the error was below the initial goal of 20 grams. However, further improvements and evaluation are needed prior launching a system able to meet the inter- and intracultural eating habits.

Functionally Optimized Neuritogenic Farinosone C Analogs: SAR-Study and Investigations on Their Mode of Action

Several natural products derived from entomopathogenic fungi have been shown to initiate neuronal differentiation in the rat pheochromocytoma PC12 cell line. After the successful completion of the total synthesis program, the reduction of structural complexity while retaining biological activity was targeted. In this study, farinosone C served as a lead structure and inspired the preparation of small molecules with reduced complexity, of which several were able to induce neurite outgrowth. This allowed for the elaboration of a detailed structure-activity relationship. Investigations on the mode of action utilizing a computational similarity ensemble approach suggested the involvement of the endocannabinoid system as potential target for our analogs and also led to the discovery of four potent new endocannabinoid transport inhibitors.

Endocannabinoid transport revisited.

Endocannabinoids are arachidonic acid-derived endogenous lipids that activate the endocannabinoid system which plays a major role in health and disease. The primary endocannabinoids are anandamide (AEA, N-arachidonoylethanolamine) and 2-arachidonoyl glycerol. While their biosynthesis and metabolism have been studied in detail, it remains unclear how endocannabinoids are transported across the cell membrane. In this review, we critically discuss the different models of endocannabinoid trafficking, focusing on AEA cellular uptake which is best studied. The evolution of the current knowledge obtained with different AEA transport inhibitors is reviewed and the confusions caused by the lack of their specificity discussed. A comparative summary of the most important AEA uptake inhibitors and the studies involving their use is provided. Based on a comprehensive literature analysis, we propose a model of facilitated AEA membrane transport followed by intracellular shuttling and sequestration. We conclude that novel and more specific probes will be essential to identify the missing targets involved in endocannabinoid membrane transport.

Localization and production of peptide endocannabinoids in the rodent CNS and adrenal medulla.

The endocannabinoid system (ECS) comprises the cannabinoid receptors CB1 and CB2 and their endogenous arachidonic acid-derived agonists 2-arachidonoyl glycerol and anandamide, which play important neuromodulatory roles. Recently, a novel class of negative allosteric CB1 receptor peptide ligands, hemopressin-like peptides derived from alpha hemoglobin, has been described, with yet unknown origin and function in the CNS. Using monoclonal antibodies we now identified the localization of RVD-hemopressin (pepcan-12) and N-terminally extended peptide endocannabinoids (pepcans) in the CNS and determined their neuronal origin. Immunohistochemical analyses in rodents revealed distinctive and specific staining in major groups of noradrenergic neurons, including the locus coeruleus (LC), A1, A5 and A7 neurons, which appear to be major sites of production/release in the CNS. No staining was detected in dopaminergic neurons. Peptidergic axons were seen throughout the brain (notably hippocampus and cerebral cortex) and spinal cord, indicative of anterograde axonal transport of pepcans. Intriguingly, the chromaffin cells in the adrenal medulla were also strongly stained for pepcans. We found specific co-expression of pepcans with galanin, both in the LC and adrenal gland. Using LC-MS/MS, pepcan-12 was only detected in non-perfused brain (∼40 pmol/g), suggesting that in the CNS it is secreted and present in extracellular compartments. In adrenal glands, significantly more pepcan-12 (400-700 pmol/g) was measured in both non-perfused and perfused tissue. Thus, chromaffin cells may be a major production site of pepcan-12 found in blood. These data uncover important areas of peptide endocannabinoid occurrence with exclusive noradrenergic immunohistochemical staining, opening new doors to investigate their potential physiological function in the ECS. This article is part of a Special Issue entitled 'Fluorescent Neuro-Ligands'.