Uncertainty Estimation in Unsupervised MR-CT Synthesis of Scoliotic Spines.

Uncertainty estimations through approximate Bayesian inference provide interesting insights to deep neural networks' behavior. In unsupervised learning tasks, where expert labels are unavailable, it becomes ever more important to critique the model through uncertainties. This paper presents a proof-of-concept for generalizing the aleatoric and epistemic uncertainties in unsupervised MR-CT synthesis of scoliotic spines. A novel adaptation of the cycle-consistency constraint in CycleGAN is proposed such that the model predicts the aleatoric uncertainty maps in addition to the standard volume-to-volume translation between Magnetic Resonance (MR) and Computed Tomography (CT) data. Ablation experiments were performed to understand uncertainty estimation as an implicit regularizer and a measure of the model's confidence. The aleatoric uncertainty helps in distinguishing between the bone and soft-tissue regions in CT and MR data during translation, while the epistemic uncertainty provides interpretable information to the user for downstream tasks.

Automatic 3D Lamina Curve Extraction from Freehand 3D Ultrasound Data using Sequential Localization Recurrent Convolutional Networks.

Freehand 3D ultrasound imaging is emerging as a promising modality for regular spine exams due to its non-invasiveness and affordability. The laminae landmarks play a critical role in depicting the 3D shape of the spine. However, the extraction of the 3D lamina curves from transverse ultrasound sequences presents a challenging task, primarily attributed to the presence of diverse contrast variations, imaging artifacts, the complex surface of vertebral bones, and the difficulties associated with probe manipulation. This paper proposes Sequential Localization Recurrent Convolutional Networks (SL-RCN), a novel deep learning model that takes the contextual relationships into account and embeds the transformation matrix feature as a 3D knowledge base to enhance accurate ultrasound sequence analysis. The assessment involved the analysis of 3D ultrasound sequences obtained from 10 healthy adult human participants, covering both the lumbar and thoracic regions. The performance of SL-RCN is evaluated through 7-fold cross-validation, employing the leave-one-participant-out strategy. The validity of the AI model training is assessed on test data from 3 participants. Normalized Discrete Fréchet Distance (NDFD) is employed as the main metric to evaluate the disparity of the extracted 3D lamina curves. In contrast to our previous 2D image analysis method, SL-RCN generates reduced left/right mean distance errors from 1.62/1.63mm to 1.41/1.40mm, and NDFDs from 0.5910/0.6389 to 0.4276/0.4567. The increase in the mean NDFD value from 7-fold cross-validation to the test-data experiment is less than 0.05. The experiments demonstrate the SL-RCN's capability in extracting accurate paired smooth lamina landmark curves.

Changes in trunk appearance following surgical correction of adolescent idiopathic scoliosis.

PURPOSE: To assess the postoperative appearance of the trunk in surgically treated scoliosis patients after a 2 year follow-up using reliable indices and compare the results with 6-month follow-up. METHODS: Forty-six Adolescent Idiopathic Scoliosis (AIS) patients (female; preop mean age 14.4 ± 2.4 years) who underwent a posterior spinal fusion from 2009 to 2018 were included in this study. All had Lenke 1A thoracic curves, with surface topography taken preoperatively, 6 months and 2 years postoperatively. To assess spinal deformity, we measured the proximal thoracic, main thoracic and thoracolumbar/lumbar Cobb angles in the frontal plane from spinal X-rays and inclinometer angles in the thoracic and lumbar regions. To assess trunk deformity, Back Surface Rotation (BSR) and Trunk Lateral Shift (TLS) were computed along the trunk. We analysed the effect of age, height, weight, Cobb angle, length of follow-up, and surgical technique. We also compared correction rates (CRs) of the spinal and trunk measurements after 6 months and 2 years. RESULTS: Good spinal correction was achieved, with Cobb angles decreasing in the whole cohort. CRs for TLS and BSR were positive (denoting improvement) for 76% and 48% of patients, respectively, after 2 years. Compared with 6 months, the mean TLS CR increased while there was no improvement for BSR on average. We found no significant association after 2 years between truncal index CRs and clinical variables (age, height, weight, preoperative Cobb angles) or surgical technique. However, there were significant correlations between the CRs of TLS and the main thoracic Cobb angle (r = 0.35), and between the CRs of BSR and thoracic inclinometer angle. CONCLUSION: Although more than 55% of the TLS was corrected after 2 years of follow-up, the BSR remained stable over time and the persistence of rib hump on the back surface could be observed. LEVEL OF EVIDENCE: III.

Denoising OCT videos based on temporal redundancy.

The identification of eye diseases and their progression often relies on a clear visualization of the anatomy and on different metrics extracted from Optical Coherence Tomography (OCT) B-scans. However, speckle noise hinders the quality of rapid OCT imaging, hampering the extraction and reliability of biomarkers that require time series. By synchronizing the acquisition of OCT images with the timing of the cardiac pulse, we transform a low-quality OCT video into a clear version by phase-wrapping each frame to the heart pulsation and averaging frames that correspond to the same instant in the cardiac cycle. Here, we compare the performance of our one-cycle denoising strategy with a deep-learning architecture, Noise2Noise, as well as classical denoising methods such as BM3D and Non-Local Means (NLM). We systematically analyze different image quality descriptors as well as region-specific metrics to assess the denoising performance based on the anatomy of the eye. The one-cycle method achieves the highest denoising performance, increases image quality and preserves the high-resolution structures within the eye tissues. The proposed workflow can be readily implemented in a clinical setting.

Automatic assessment of scoliosis surgery outcome on trunk shape using left-right trunk asymmetry.

PURPOSE: To present a novel set of Left-Right Trunk Asymmetry (LRTA) indices and use them to assess the postoperative appearance of the trunk in Adolescent Idiopathic Scoliosis (AIS) patients. METHODS: We hypothesize that LRTA measurements provide complementary information to existing trunk asymmetry indices when documenting the outcome of scoliosis surgery. Forty-nine AIS patients with thoracic curves who underwent posterior spinal fusion were included. All had surface topography scans taken preoperatively and at least 6 months postoperatively. We documented spinal curvature using Radiographic Cobb angles, scoliometer readings and coronal balance. To evaluate Global Trunk Asymmetry (GTA), we used the standard measures of Back Surface Rotation (BSR) and Trunk Lateral Shift (TLS). To measure LRTA, we identified asymmetry areas as regions of significant deviation between the left and right sides of the 3D back surface. New parameters called Deformation Rate (DR) and Maximum Asymmetry (MA) were measured in different regions based on the asymmetry areas. We compared the GTA and LRTA changes with those in spinal curvature before and after surgery. RESULTS: The GTA indices, mainly TLS, showed improvement for more than 75% of patients. There was significant improvement of LRTA in the shoulder blades and waist regions (95% and 80% of patients respectively). CONCLUSION: We report positive outcomes for LRTA in the majority of patients, specifically in the shoulder blades and waist, even when no reduction of BSR is observed. The proposed indices can evaluate local trunk asymmetries and the degree to which they are improved or worsened after scoliosis surgery.

Joint classification and segmentation for an interpretable diagnosis of acute respiratory distress syndrome from chest x-rays.

Purpose: Acute respiratory distress syndrome (ARDS) is a life-threatening condition that can cause a dramatic drop in blood oxygen levels due to widespread lung inflammation. Chest radiography is widely used as a primary modality to detect ARDS due to its crucial role in diagnosing the syndrome, and the x-ray images can be obtained promptly. However, despite the extensive literature on chest x-ray (CXR) image analysis, there is limited research on ARDS diagnosis due to the scarcity of ARDS-labeled datasets. Additionally, many machine learning-based approaches result in high performance in pulmonary disease diagnosis, but their decisions are often not easily interpretable, which can hinder their clinical acceptance. This work aims to develop a method for detecting signs of ARDS in CXR images that can be clinically interpretable. Approach: To achieve this goal, an ARDS-labeled dataset of chest radiography images is gathered and annotated for training and evaluation of the proposed approach. The proposed deep classification-segmentation model, Dense-Ynet, provides an interpretable framework for automatically diagnosing ARDS in CXR images. The model takes advantage of lung segmentation in diagnosing ARDS. By definition, ARDS causes bilateral diffuse infiltrates throughout the lungs. To consider the local involvement of lung areas, each lung is divided into upper and lower halves, and our model classifies the resulting lung quadrants. Results: The quadrant-based classification strategy yields the area under the receiver operating characteristic curve of 95.1% (95% CI 93.5 to 96.1), which allows for providing a reference for the model's predictions. In terms of segmentation, the model accurately identifies lung regions in CXR images even when lung boundaries are unclear in abnormal images. Conclusions: This study provides an interpretable decision system for diagnosing ARDS, by following the definition used by clinicians for the diagnosis of ARDS from CXR images.

Evaluation of the SIMULRESP: A simulation software of child and teenager cardiorespiratory physiology.

BACKGROUND: Mathematical models based on the physiology when programmed as a software can be used to teach cardiorespiratory physiology and to forecast the effect of various ventilatory support strategies. We developed a cardiorespiratory simulator for children called "SimulResp." The purpose of this study was to evaluate the quality of SimulResp. METHODS: SimulResp quality was evaluated on accuracy, robustness, repeatability, and reproducibility. Blood gas values (pH, PaCO2 , PaO2,  and SaO2 ) were simulated for several subjects with different characteristics and in different situations and compared to expected values available as reference. The correlation between reference and simulated data was evaluated by the coefficient of determination and Intraclass correlation coefficient. The agreement was evaluated with the Bland & Altman analysis. RESULTS: SimulResp produced healthy child physiological values within normal range (pH 7.40 ± 0.5; PaCO2 40 ± 5 mmHg; PaO2 90 ± 10 mmHg; SaO2 97 ± 3%) starting from a weight of 25-35 kg, regardless of ventilator support. SimulResp failed to simulate accurate values for subjects under 25 kg and/or affected with pulmonary disease and mechanically ventilated. Based on the repeatability was considered as excellent and the reproducibility as mild to good. SimulResp's prediction remains stable within time. CONCLUSIONS: The cardiorespiratory simulator SimulResp requires further development before future integration into a clinical decision support system.

A Web-Based Platform for the Automatic Stratification of ARDS Severity.

Acute respiratory distress syndrome (ARDS), including severe pulmonary COVID infection, is associated with a high mortality rate. It is crucial to detect ARDS early, as a late diagnosis may lead to serious complications in treatment. One of the challenges in ARDS diagnosis is chest X-ray (CXR) interpretation. ARDS causes diffuse infiltrates through the lungs that must be identified using chest radiography. In this paper, we present a web-based platform leveraging artificial intelligence (AI) to automatically assess pediatric ARDS (PARDS) using CXR images. Our system computes a severity score to identify and grade ARDS in CXR images. Moreover, the platform provides an image highlighting the lung fields, which can be utilized for prospective AI-based systems. A deep learning (DL) approach is employed to analyze the input data. A novel DL model, named Dense-Ynet, is trained using a CXR dataset in which clinical specialists previously labelled the two halves (upper and lower) of each lung. The assessment results show that our platform achieves a recall rate of 95.25% and a precision of 88.02%. The web platform, named PARDS-CxR, assigns severity scores to input CXR images that are compatible with current definitions of ARDS and PARDS. Once it has undergone external validation, PARDS-CxR will serve as an essential component in a clinical AI framework for diagnosing ARDS.

Mapping Pulsatile Optic Nerve Head Deformation Using OCT.

Objective: To develop a noninvasive technique to quantitatively assess the pulsatile deformation due to cardiac contractions of the optic nerve head (ONH). Design: Evaluation of a diagnostic test or technology. Participants: Healthy subjects with no history of refractive surgery, divided into 2 cohorts on the basis of their axial length (AL). Methods: We present a noninvasive technique to quantitatively assess the pulsatile deformation of the ONH tissue by combining high-frequency OCT imaging and widely available image processing algorithms. We performed a thorough validation of the approach, numerically and experimentally, evaluating the sensitivity of the method to artificially induced deformation and its robustness to different noise levels. We performed deformation measurements in cohorts of healthy (n = 9) and myopic (n = 5) subjects in different physiological strain conditions by calculating the amplitude of tissue displacement in both the primary position and abduction. The head rotation was measured using a goniometer. During imaging in abduction, the head was rotated 40° ± 3°, and subjects were instructed to direct their gaze toward the OCT visual target. Main Outcome Measures: Pulsatile tissue displacement maps. Results: The robustness of the method was assessed using artificial deformations and increasing noise levels. The results show acceptable absolute errors before the noise simulations grossly exaggerate image degradation. For the group of subjects with AL of < 25 mm (n = 9), the median pulsatile displacement of the ONH was 7.8 ± 1.3 µm in the primary position and 8.9 ± 1.2 µm in abduction. The Wilcoxon test showed a significant difference (P ≤ 0.005) between the 2 paired measures. Reproducibility was tested in 2 different sessions in 5 different subjects with the same intraocular pressure, and an intraclass correlation coefficient of 0.99 was obtained (P < 0.005). Conclusions: The computational pipeline demonstrated good reproducibility and had the capacity to accurately map the pulsatile deformation of the optic nerve. In a clinical setting, we detected physiological changes in normal subjects supporting its translation potential as a novel biomarker for the diagnosis and progression of optic nerve diseases.

Focused Attention in Transformers for interpretable classification of retinal images.

Vision Transformers have recently emerged as a competitive architecture in image classification. The tremendous popularity of this model and its variants comes from its high performance and its ability to produce interpretable predictions. However, both of these characteristics remain to be assessed in depth on retinal images. This study proposes a thorough performance evaluation of several Transformers compared to traditional Convolutional Neural Network (CNN) models for retinal disease classification. Special attention is given to multi-modality imaging (fundus and OCT) and generalization to external data. In addition, we propose a novel mechanism to generate interpretable predictions via attribution maps. Existing attribution methods from Transformer models have the disadvantage of producing low-resolution heatmaps. Our contribution, called Focused Attention, uses iterative conditional patch resampling to tackle this issue. By means of a survey involving four retinal specialists, we validated both the superior interpretability of Vision Transformers compared to the attribution maps produced from CNNs and the relevance of Focused Attention as a lesion detector.

[Automatic extraction of vertebral landmarks from ultrasound images]. / Extraction automatique de repères vertébraux à partir d'échographies.

Some forms of myopathies such as Duchenne muscular dystrophy cause a progressive degeneration of the patient's muscles. This results in the development of scoliosis, which increases in severity over time. The clinical standard for monitoring scoliosis is to perform an X-ray on a regular basis. Unfortunately, repeated exposure to X-rays is harmful to the patient's health. Ultrasound imaging is a radiation-free modality that uses ultrasound (US) waves. However, the interpretation of vertebral ultrasound images is often difficult due to the variable quality of the image. In order to tackle this challenge, we present a method to localize the vertebrae on US images automatically. The validation of this reproducible approach suggests that it would be possible, in the long term, to replace part of the X-ray exams by US imaging.

TITLE: Extraction automatique de repères vertébraux à partir d'échographies. ABSTRACT: Certaines formes de myopathies telles que la dystrophie musculaire de Duchenne entraînent une dégénérescence progressive des muscles chez le patient. Ceci se traduit par l'apparition d'une scoliose dont la gravité augmente au cours du temps. La norme clinique pour le suivi de la scoliose consiste à réaliser un examen radiographique. Malheureusement, l'exposition répétée aux rayons X est nocive pour la santé du patient. L'échographie est une technique d'imagerie médicale non irradiante qui utilise des ondes ultrasonores (US). Cependant, l'interprétation des échographies de vertèbres est souvent difficile en raison de la qualité variable des images. En réponse à ce défi, nous présentons une méthode pour localiser automatiquement les vertèbres sur les échographies. La validation de cette approche reproductible laisse à penser qu'il serait possible, à terme, de remplacer une partie des examens radiographiques standards par l'échographie.

Detection of doxorubicin-induced cardiotoxicity using myocardial T1 and T2 relaxation times in childhood acute lymphoblastic leukemia survivors.

Doxorubicin leads to dose-dependent cardiotoxicity in childhood acute lymphoblastic leukemia (ALL) survivors. The first aim was to propose a contour-based estimation of T1 and T2 relaxation times based on the myocardial area, while our second aim was to evaluate native T1, post-gadolinium T1 and T2 relaxation time sensitivity to detect myocardial changes. A total of 84 childhood ALL survivors were stratified in regard to their prognostic risk groups: standard risk (SR), n = 20), high-risk with and without dexrazoxane (HR + DEX, n = 39 and HR, n = 25). Survivors' mean age was of 22.0 ± 6.9 years, with a mean age at cancer diagnosis of 8.0 ± 5.2 years. CMR acquisitions were performed on a 3 T MRI system and included an ECG-gated 3(3)3(3)5 MOLLI sequence for T1 mapping and an ECG-gated T2-prepared TrueFISP sequence for T2 mapping. Myocardial contours were semi-automatically segmented using an interactive implementation of cubic Bezier curves. We found excellent repeatability between operators for native T1 (ICC = 0.91), and good repeatability between operators for post-gadolinium T1 (ICC = 0.84) and T2 (ICC = 0.79). Bland and Altman tests demonstrated a strong agreement between our contour-based method and images analyzed using the CVI42 software on the measure of native T1, post-gadolinium T1, and T2. No significant differences between survivors' prognostic risk groups in native T1 were reported, while we observed significant differences between survivors' prognostic risk groups in post-gadolinium T1 and T2. Significant differences were observed between male and female survivors. Differences between groups were also observed in partition coefficients, but no significant differences were observed between male and female survivors. The use of CMR parameters with native T1, post-gadolinium T1, and T2 allowed to show that survivors at a high-risk prognostic were more exposed to doxorubicin-related cardiotoxicity than those who were at a standard risk prognostic or who received dexrazoxane treatments.

A simulation study investigating potential diffusion-based MRI signatures of microstrokes.

Recent studies suggested that cerebrovascular micro-occlusions, i.e. microstokes, could lead to ischemic tissue infarctions and cognitive deficits. Due to their small size, identifying measurable biomarkers of these microvascular lesions remains a major challenge. This work aims to simulate potential MRI signatures combining arterial spin labeling (ASL) and multi-directional diffusion-weighted imaging (DWI). Driving our hypothesis are recent observations demonstrating a radial reorientation of microvasculature around the micro-infarction locus during recovery in mice. Synthetic capillary beds, randomly- and radially-oriented, and optical coherence tomography (OCT) angiograms, acquired in the barrel cortex of mice (n = 5) before and after inducing targeted photothrombosis, were analyzed. Computational vascular graphs combined with a 3D Monte-Carlo simulator were used to characterize the magnetic resonance (MR) response, encompassing the effects of magnetic field perturbations caused by deoxyhemoglobin, and the advection and diffusion of the nuclear spins. We quantified the minimal intravoxel signal loss ratio when applying multiple gradient directions, at varying sequence parameters with and without ASL. With ASL, our results demonstrate a significant difference (p < 0.05) between the signal-ratios computed at baseline and 3 weeks after photothrombosis. The statistical power further increased (p < 0.005) using angiograms measured at week 4. Without ASL, no reliable signal change was found. We found that higher ratios, and accordingly improved significance, were achieved at lower magnetic field strengths (e.g., B0 = 3T) and shorter echo time TE (< 16 ms). Our simulations suggest that microstrokes might be characterized through ASL-DWI sequence, providing necessary insights for posterior experimental validations, and ultimately, future translational trials.

Atlas-based score for automatic glaucoma risk stratification.

Glaucoma is a disease that affects the optic nerve and can lead to blindness. The cup-to-disc ratio (CDR) measurement is one of the key clinical indicators for glaucoma assessment. However, the CDR only evaluates the relative sizes of the cup and optic disc (OD) via their diameters, and does not characterize local morphological changes that can inform clinicians on early signs of glaucoma. In this work, we propose a novel glaucoma score based on a statistical atlas framework that automatically quantifies the deformations of the OD region induced by glaucoma. A deep-learning approach is first used to segment the optic cup with a dedicated atlas-based data augmentation strategy. The segmented OD region (disc, cup and vessels) is then registered to the statistical OD atlas and the deformation is projected onto the atlas eigenvectors. The atlas glaucoma score (AGS) is then obtained by a linear combination of the principal modes of deformation of the atlas with linear discriminant analysis. The AGS performs better than the CDR on the three datasets used for evaluation, including RIM-ONE and ORIGA650. Compared to the CDR measurement, which yields an area under the ROC curve (AUC) of 91.4% using the expert segmentations, the AGS achieves an AUC of 98.2%. Our novel glaucoma score captures more complex deformations within the optic disc region than the CDR can. Such morphological changes are the first cue of glaucoma onset, before the visual field is affected. The proposed approach can thus significantly improve early detection of glaucoma.

Laplacian Flow Dynamics on Geometric Graphs for Anatomical Modeling of Cerebrovascular Networks.

Generating computational anatomical models of cerebrovascular networks is vital for improving clinical practice and understanding brain oxygen transport. This is achieved by extracting graph-based representations based on pre-mapping of vascular structures. Recent graphing methods can provide smooth vessels trajectories and well-connected vascular topology. However, they require water-tight surface meshes as inputs. Furthermore, adding vessels radii information on their graph compartments restricts their alignment along vascular centerlines. Here, we propose a novel graphing scheme that works with relaxed input requirements and intrinsically captures vessel radii information. The proposed approach is based on deforming geometric graphs constructed within vascular boundaries. Under a laplacian optimization framework, we assign affinity weights on the initial geometry that drives its iterative contraction toward vessels centerlines. We present a mechanism to decimate graph structure at each run and a convergence criterion to stop the process. A refinement technique is then introduced to obtain final vascular models. Our implementation is available on https://github.com/Damseh/VascularGraph. We benchmarked our results with that obtained using other efficient and state-of-the-art graphing schemes, validating on both synthetic and real angiograms acquired with different imaging modalities. The experiments indicate that the proposed scheme produces the lowest geometric and topological error rates on various angiograms. Furthermore, it surpasses other techniques in providing representative models that capture all anatomical aspects of vascular structures.

End-to-End Deep Learning Model for Cardiac Cycle Synchronization from Multi-View Angiographic Sequences.

Dynamic reconstructions (3D+T) of coronary arteries could give important perfusion details to clinicians. Temporal matching of the different views, which may not be acquired simultaneously, is a prerequisite for an accurate stereo-matching of the coronary segments. In this paper, we show how a neural network can be trained from angiographic sequences to synchronize different views during the cardiac cycle using raw x-ray angiography videos exclusively. First, we train a neural network model with angiographic sequences to extract features describing the progression of the cardiac cycle. Then, we compute the distance between the feature vectors of every frame from the first view with those from the second view to generate distance maps that display stripe patterns. Using pathfinding, we extract the best temporally coherent associations between each frame of both videos. Finally, we compare the synchronized frames of an evaluation set with the ECG signals to show an alignment with 96.04% accuracy.

Dense-Unet: a light model for lung fields segmentation in Chest X-Ray images.

Automatic and accurate lung segmentation in chest X-ray (CXR) images is fundamental for computer-aided diagnosis systems since the lung is the region of interest in many diseases and also it can reveal useful information by its contours. While deep learning models have reached high performances in the segmentation of anatomical structures, the large number of training parameters is a concern since it increases memory usage and reduces the generalization of the model. To address this, a deep CNN model called Dense-Unet is proposed in which, by dense connectivity between various layers, information flow increases throughout the network. This lets us design a network with significantly fewer parameters while keeping the segmentation robust. To the best of our knowledge, Dense-Unet is the lightest deep model proposed for the segmentation of lung fields in CXR images. The model is evaluated on the JSRT and Montgomery datasets and experiments show that the performance of the proposed model is comparable with state-of-the-art methods.

OrgaNet: A Robust Network for Subcellular Organelles Classification in Fluorescence Microscopy Images.

Automatic identification of subcellular compartments of proteins in fluorescence microscopy images is an important task to quantitatively evaluate cellular processes. A common problem for the development of deep learning based classifiers is that there is only a limited number of labeled images available for training. To address this challenge, we propose a new approach for subcellular organelles classification combining an effective and efficient architecture based on a compact Convolutional Neural Network and deep embedded clustering algorithm. We validate our approach on a benchmark of HeLa cell microscopy images. The network both yields high accuracy that outperforms state of the art methods and has significantly small number of parameters. More interestingly, experimental results show that our method is strongly robust against limited labeled data for training, requiring four times less annotated data than usual while maintaining the high accuracy of 93.9%.

Automated Analysis of Brain Microvasculature: From Segmentation to Anatomical Modeling.

Two-photon microscopy (TPM) can provide a detailed microscopic information of cerebrovascular structures. Extracting anatomical vascular models from TPM angiograms remains a tedious task due to image degeneration associated with TPM acquisitions and the complexity of microvascular networks. Here, we propose a fully automated pipeline capable of providing useful anatomical models of vascular structures captured with TPM. In the proposed method, we segment blood vessels using a fully convolutional neural network and employ the resulting binary labels to create an initial geometric graph enclosed within vessels boundaries. The initial geometry is then decimated and refined to form graphed curve skeletons that can retain both the vascular shape and its topology. We validate the proposed method on 3D realistic TPM angiographies and compare our results with that obtained through manual annotations.

3D reconstruction of the human trunk for designing personalized braces : Precision study.

The customized design of braces for adolescent idiopathic scoliosis (AIS) treatment requires the acquisition of the 3D external geometry of the patients' trunks. Three body scanning systems are available at CHU Sainte-Justine in Montreal: a fixed system of InSpeck Capturor II LF digitizers and two portable scanners, BodyScan and Structure Sensor. The aim of this study is to compare them by evaluating their accuracy and repeatability. To achieve this, we placed 46 surface markers on an anthropomorphic manikin and scanned it three times with each system. We also measured the 3D coordinates of the same markers using a coordinate measuring machine (CMM), serving as ground-truth. We evaluated the repeatability and accuracy of the three systems: the former, by measuring the bidirectional mean distance between the three surfaces acquired with a given modality; the latter, by calculating the residual normal distance separating each of the 3D surfaces and the CMM point cloud. We also compared texture mapping accuracy between InSpeck and Structure Sensor by examining the CMM point cloud versus the marker 3D coordinates selected on the trunk surface. The results show good accuracy and repeatability for all three systems, with slightly better geometric accuracy for BodyScan (p-value ≈ 10-6). In terms of texture mapping, InSpeck showed better accuracy than Structure Sensor (p-value = 0.0059).