Automatic quantification of scapular and glenoid morphology from CT scans using deep learning.

OBJECTIVES: To develop and validate an open-source deep learning model for automatically quantifying scapular and glenoid morphology using CT images of normal subjects and patients with glenohumeral osteoarthritis. MATERIALS AND METHODS: First, we used deep learning to segment the scapula from CT images and then to identify the location of 13 landmarks on the scapula, 9 of them to establish a coordinate system unaffected by osteoarthritis-related changes, and the remaining 4 landmarks on the glenoid cavity to determine the glenoid size and orientation in this scapular coordinate system. The glenoid version, glenoid inclination, critical shoulder angle, glenopolar angle, glenoid height, and glenoid width were subsequently measured in this coordinate system. A 5-fold cross-validation was performed to evaluate the performance of this approach on 60 normal/non-osteoarthritic and 56 pathological/osteoarthritic scapulae. RESULTS: The Dice similarity coefficient between manual and automatic scapular segmentations exceeded 0.97 in both normal and pathological cases. The average error in automatic scapular and glenoid landmark positioning ranged between 1 and 2.5 mm and was comparable between the automatic method and human raters. The automatic method provided acceptable estimates of glenoid version (R2 = 0.95), glenoid inclination (R2 = 0.93), critical shoulder angle (R2 = 0.95), glenopolar angle (R2 = 0.90), glenoid height (R2 = 0.88) and width (R2 = 0.94). However, a significant difference was found for glenoid inclination between manual and automatic measurements (p < 0.001). CONCLUSIONS: This open-source deep learning model enables the automatic quantification of scapular and glenoid morphology from CT scans of patients with glenohumeral osteoarthritis, with sufficient accuracy for clinical use.

A Multidisciplinary Hyper-Modeling Scheme in Personalized In Silico Oncology: Coupling Cell Kinetics with Metabolism, Signaling Networks, and Biomechanics as Plug-In Component Models of a Cancer Digital Twin.

The massive amount of human biological, imaging, and clinical data produced by multiple and diverse sources necessitates integrative modeling approaches able to summarize all this information into answers to specific clinical questions. In this paper, we present a hypermodeling scheme able to combine models of diverse cancer aspects regardless of their underlying method or scale. Describing tissue-scale cancer cell proliferation, biomechanical tumor growth, nutrient transport, genomic-scale aberrant cancer cell metabolism, and cell-signaling pathways that regulate the cellular response to therapy, the hypermodel integrates mutation, miRNA expression, imaging, and clinical data. The constituting hypomodels, as well as their orchestration and links, are described. Two specific cancer types, Wilms tumor (nephroblastoma) and non-small cell lung cancer, are addressed as proof-of-concept study cases. Personalized simulations of the actual anatomy of a patient have been conducted. The hypermodel has also been applied to predict tumor control after radiotherapy and the relationship between tumor proliferative activity and response to neoadjuvant chemotherapy. Our innovative hypermodel holds promise as a digital twin-based clinical decision support system and as the core of future in silico trial platforms, although additional retrospective adaptation and validation are necessary.

Impact of Tissue Damage and Hemodynamics on Restenosis Following Percutaneous Transluminal Angioplasty: A Patient-Specific Multiscale Model.

Multiscale agent-based modeling frameworks have recently emerged as promising mechanobiological models to capture the interplay between biomechanical forces, cellular behavior, and molecular pathways underlying restenosis following percutaneous transluminal angioplasty (PTA). However, their applications are mainly limited to idealized scenarios. Herein, a multiscale agent-based modeling framework for investigating restenosis following PTA in a patient-specific superficial femoral artery (SFA) is proposed. The framework replicates the 2-month arterial wall remodeling in response to the PTA-induced injury and altered hemodynamics, by combining three modules: (i) the PTA module, consisting in a finite element structural mechanics simulation of PTA, featuring anisotropic hyperelastic material models coupled with a damage formulation for fibrous soft tissue and the element deletion strategy, providing the arterial wall damage and post-intervention configuration, (ii) the hemodynamics module, quantifying the post-intervention hemodynamics through computational fluid dynamics simulations, and (iii) the tissue remodeling module, based on an agent-based model of cellular dynamics. Two scenarios were explored, considering balloon expansion diameters of 5.2 and 6.2 mm. The framework captured PTA-induced arterial tissue lacerations and the post-PTA arterial wall remodeling. This remodeling process involved rapid cellular migration to the PTA-damaged regions, exacerbated cell proliferation and extracellular matrix production, resulting in lumen area reduction up to 1-month follow-up. After this initial reduction, the growth stabilized, due to the resolution of the inflammatory state and changes in hemodynamics. The similarity of the obtained results to clinical observations in treated SFAs suggests the potential of the framework for capturing patient-specific mechanobiological events occurring after PTA intervention.

Predicted vs. measured paraspinal muscle activity in adolescent idiopathic scoliosis patients: EMG validation of optimization-based musculoskeletal simulations.

Musculoskeletal (MSK) models offer great potential for predicting the muscle forces required to inform more detailed simulations of vertebral endplate loading in adolescent idiopathic scoliosis (AIS). In this work, simulations based on static optimization were compared with in vivo measurements in two AIS patients to determine whether computational approaches alone are sufficient for accurate prediction of paraspinal muscle activity during functional activities. We used biplanar radiographs and marker-based motion capture, ground reaction force, and electromyography (EMG) data from two patients with mild and moderate thoracolumbar AIS (Cobb angles: 21° and 45°, respectively) during standing while holding two weights in front (reference position), walking, running, and object lifting. Using a fully automated approach, 3D spinal shape was extracted from the radiographs. Geometrically personalized OpenSim-based MSK models were created by deforming the spine of pre-scaled full-body models of children/adolescents. Simulations were performed using an experimentally controlled backward approach. Differences between model predictions and EMG measurements of paraspinal muscle activity (both expressed as a percentage of the reference position values) at three different locations around the scoliotic main curve were quantified by root mean square error (RMSE) and cross-correlation (XCorr). Predicted and measured muscle activity correlated best for mild AIS during object lifting (XCorr's ≥ 0.97), with relatively low RMSE values. For moderate AIS as well as the walking and running activities, agreement was lower, with XCorr reaching values of 0.51 and comparably high RMSE values. This study demonstrates that static optimization alone seems not appropriate for predicting muscle activity in AIS patients, particularly in those with more than mild deformations as well as when performing upright activities such as walking and running.

Glenohumeral joint force prediction with deep learning.

Deep learning models (DLM) are efficient replacements for computationally intensive optimization techniques. Musculoskeletal models (MSM) typically involve resource-intensive optimization processes for determining joint and muscle forces. Consequently, DLM could predict MSM results and reduce computational costs. Within the total shoulder arthroplasty (TSA) domain, the glenohumeral joint force represents a critical MSM outcome as it can influence joint function, joint stability, and implant durability. Here, we aimed to employ deep learning techniques to predict both the magnitude and direction of the glenohumeral joint force. To achieve this, 959 virtual subjects were generated using the Markov-Chain Monte-Carlo method, providing patient-specific parameters from an existing clinical registry. A DLM was constructed to predict the glenohumeral joint force components within the scapula coordinate system for the generated subjects with a coefficient of determination of 0.97, 0.98, and 0.98 for the three components of the glenohumeral joint force. The corresponding mean absolute errors were 11.1, 12.2, and 15.0 N, which were about 2% of the maximum glenohumeral joint force. In conclusion, DLM maintains a comparable level of reliability in glenohumeral joint force estimation with MSM, while drastically reducing the computational costs.

Depth-dependent mechanical properties of the human cornea by uniaxial extension.

The purpose of this study was to investigate the depth-dependent biomechanical properties of the human corneal stroma under uniaxial tensile loading. Human stroma samples were obtained after the removal of Descemet's membrane in the course of Descemet's membrane endothelial keratoplasty (DMEK) transplantation. Uniaxial tensile tests were performed at three different depths: anterior, central, and posterior on 2 × 6 × 0.15 mm strips taken from the central DMEK graft. The measured force-displacement data were used to calculate stress-strain curves and to derive the tangent modulus. The study showed that mechanical strength decreased significantly with depth. The anterior cornea appeared to be the stiffest, with a stiffness approximately 18% higher than that of the central cornea and approximately 38% higher than that of the posterior layer. Larger variations in mechanical response were observed in the posterior group, probably due to the higher degree of alignment of the collagen fibers in the posterior sections of the cornea. This study contributes to a better understanding of the biomechanical tensile properties of the cornea, which has important implications for the development of new treatment strategies for corneal diseases. Accurate quantification of tensile strength as a function of depth is critical information that is lacking in human corneal biomechanics to develop numerical models and new treatment methods.

Optomechanical assessment of photorefractive corneal cross-linking via optical coherence elastography.

Purpose: Corneal cross-linking (CXL) has recently been used with promising results to positively affect corneal refractive power in the treatment of hyperopia and mild myopia. However, understanding and predicting the optomechanical changes induced by this procedure are challenging. Methods: We applied ambient pressure modulation based optical coherence elastography (OCE) to quantify the refractive and mechanical effects of patterned CXL and their relationship to energy delivered during the treatment on porcine corneas. Three different patterned treatments were performed, designed according to Zernike polynomial functions (circle, astigmatism, coma). In addition, three different irradiation protocols were analyzed: standard Dresden CXL (fluence of 5.4 J/cm2), accelerated CXL (fluence of 5.4 J/cm2), and high-fluence CXL (fluence of 16.2 J/cm2). The axial strain distribution in the stroma induced by ocular inflation (Δp = 30 mmHg) was quantified, maps of the anterior sagittal curvature were constructed and cylindrical refraction was assessed. Results: Thirty minutes after CXL, there was a statistically significant increase in axial strain amplitude (p < 0.050) and a reduction in sagittal curvature (p < 0.050) in the regions treated with all irradiation patterns compared to the non-irradiated ones. Thirty-6 hours later, the non-irradiated regions showed compressive strains, while the axial strain in the CXL-treated regions was close to zero, and the reduction in sagittal curvature observed 30 minutes after the treatment was maintained. The Dresden CXL and accelerated CXL produced comparable amounts of stiffening and refractive changes (p = 0.856), while high-fluence CXL produced the strongest response in terms of axial strain (6.9‰ ± 1.9‰) and refractive correction (3.4 ± 0.9 D). Tripling the energy administered during CXL resulted in a 2.4-fold increase in the resulting refractive correction. Conclusion: OCE showed that refractive changes and alterations in corneal biomechanics are directly related. A patient-specific selection of both, the administered UV fluence and the irradiation pattern during CXL is promising to allow customized photorefractive corrections in the future.

Patient-specific finite element analysis of human corneal lenticules: An experimental and numerical study.

The number of elective refractive surgeries is constantly increasing due to the drastic increase in myopia prevalence. Since corneal biomechanics are critical to human vision, accurate modeling is essential to improve surgical planning and optimize the results of laser vision correction. In this study, we present a numerical model of the anterior cornea of young patients who are candidates for laser vision correction. Model parameters were determined from uniaxial tests performed on lenticules of patients undergoing refractive surgery by means of lenticule extraction, using patient-specific models of the lenticules. The models also took into account the known orientation of collagen fibers in the tissue, which have an isotropic distribution in the corneal plane, while they are aligned along the corneal curvature and have a low dispersion outside the corneal plane. The model was able to reproduce the experimental data well with only three parameters. These parameters, determined using a realistic fiber distribution, yielded lower values than those reported in the literature. Accurate characterization and modeling of the cornea of young patients is essential to study better refractive surgery for the population undergoing these treatments, to develop in silico models that take corneal biomechanics into account when planning refractive surgery, and to provide a basis for improving visual outcomes in the rapidly growing population undergoing these treatments.

Analysis of Biomechanical Response After Corneal Crosslinking with Different Fluence Levels in Porcine Corneas.

PURPOSE: To evaluate corneal stiffening of porcine corneas induced by corneal crosslinking (CXL) with constant irradiance as a function of total fluence. METHODS: Ninety corneas from freshly enucleated porcine eyes were divided into five groups of 18 eyes. Groups 1-4 underwent epi-off CXL using a dextran-based riboflavin solution and an irradiance of 18 mW/cm2, group 5 served as the control group. Groups 1 to 4 were treated with a total fluence of 20, 15, 10.8, and 5.4 J/cm2, respectively. Thereafter, biomechanical measurements were performed on 5 mm wide and 6 mm long strips using an uniaxial material tester. Pachymetry measurements were performed on each cornea. RESULTS: At 10% strain, the stress was 76, 56, 52, and 31% higher in groups 1-4, respectively compared to the control group. The Young's modulus was 2.85 MPa for group 1, 2.53 MPa for group 2, 2.46 MPa for group 3, 2.12 MPa for group 4, and 1.62 MPa for the control group. The difference between groups 1 to 4 and the control group 5 were statistically significant (p = <0.001; p = <0.001; p = <0.001; p = 0.021). In addition, group 1 showed significantly more stiffening than group 4 (p = <0.001), no other significant differences were found. Pachymetry measurements revealed no statistically significant differences among the five groups. CONCLUSION: Additional mechanical stiffening can be achieved by increasing the fluence of the CXL. There was no threshold detected up to 20 J/cm2. A higher fluence could compensate the weaker effect of accelerated or epi-on CXL procedures.

Age- and sex-specific normative values of bone mineral density in the adult glenoid.

The objective of this study was to determine the normative bone mineral density (BMD) of cortical and trabecular bone regions in the adult glenoid and its dependence on the subject's age and sex. We analyzed computed tomography (CT) scans of 441 shoulders (310 males, 18-69 years) without any signs of glenohumeral joint pathology. Glenoid BMD was automatically quantified in six volumes of interest (VOIs): cortical bone (CO), subchondral cortical plate (SC), subchondral trabecular bone (ST), and three adjacent layers of trabecular bone (T1, T2, and T3). BMD was measured in Hounsfield unit (HU). We evaluated the association between glenoid BMD and sex and age with the Student's t test and Pearson's correlation coefficient (r), respectively. The lambda-mu-sigma method was used to determine age- and sex-specific normative values of glenoid BMD in cortical (CO and SC) and trabecular (ST, T1, T2, and T3) bone. Glenoid BMD was higher in males than females, in most age groups and most VOIs. Before 40 years old, the effect of age on BMD was very weak in both males and females. After 40 years old, BMD declined over time in all VOIs. This BMD decline with age was greater in females (cortical: r = -0.45, trabecular: r = -0.41) than in males (cortical: r = -0.30; trabecular: r = -0.32). These normative glenoid BMD values could prove clinically relevant in the diagnosis and management of patients with various shoulder disorders, in particular glenohumeral osteoarthritis and shoulder arthroplasty or shoulder instability, as well as in related research.

Traditional T1-S1 Measurement of the Spinal Length on X-ray Images Does Not Correlate With the True Length of the Spine.

BACKGROUND: The T1-S1 distance to evaluate spinal length is traditionally measured as a straight line on an anteroposterior radiograph. However, this method may not reflect the true 3-dimensional (3D) spinal length. The objective of the study was to evaluate the difference between the traditional T1-S1 measurement and a 3D reconstruction from standard x-ray imaging. METHODS: Radiological assessment and 3D reconstruction of spinal length in pediatric patients with various spine deformities. The 3D reconstruction derived from standard biplanar spine x-ray images using a specialized but free available software and calibration device. Direct comparison of length, intraobserver variance for repeated measurements, as well as interobserver correlation for both measurement methods and between different levels of training were evaluated. Furthermore, the influence on spinal length by the degree of spinal deformity as well as other factors was analyzed. RESULTS: A total of 39 x-ray images from 35 patients at a mean age of 15.4 years (8.9-26.8 years) were evaluated. There was excellent agreement for intra- and interobserver correlation for both measurement techniques. Spinal length assessed using 3D reconstruction was significantly longer compared with the traditional T1-S1 distance, on average 2.7 cm (0.5-6.1 cm). There was also a significant positive correlation between the maximum extent of the deformity and the difference in spinal length. CONCLUSIONS: Traditional T1-S1 distance significantly underestimates the true length of the spine. A 3D measurement reflects the real length of the spine more adequately. CLINICAL RELEVANCE: Such information is relevant to the treating spine surgeon when planning or assessing therapeutic measures, especially in advanced deformities.

Orientation and depth dependent mechanical properties of the porcine cornea: Experiments and parameter identification.

The porcine cornea is a standard animal model in ophthalmic research, making its biomechanical characterization and modeling important to develop novel treatments such as crosslinking and refractive surgeries. In this study, we present a numerical model of the porcine cornea based on experimental measurements that captures both the depth dependence and orientation dependence of the mechanical response. The mechanical parameters of the established anisotropic hyperelastic material models of Gasser, Holzapfel and Ogden (HGO) and Markert were determined using tensile tests. Corneas were cut with a femtosecond laser in the anterior (100 µm), central (350 µm), and posterior (600 µm) regions into nasal-temporal, superior-inferior, and diagonal strips of 150 µm thickness. These uniformly thick strips were tested at a low speed using a single-axis testing machine. The results showed that the corneal mechanical properties remained constant in the anterior half of the cornea regardless of orientation, but that the material softened in the posterior layer. These results are consistent with the circular orientation of collagen observed in porcine corneas using X-ray scattering. In addition, the parameters obtained for the HGO model were able to reproduce the published inflation tests, indicating that it is suitable for simulating the mechanical response of the entire cornea. Such a model constitutes the basis for in silico platforms to develop new ophthalmic treatments. In this way, researchers can match their experimental surrogate porcine model with a numerical counterpart and validate the prediction of their algorithms in a complete and accessible environment.

Towards a better understanding of the posttreatment hemodynamic behaviors in femoropopliteal arteries through personalized computational models based on OCT images.

The hemodynamic behavior following endovascular treatment of patients with peripheral arterial disease plays a significant role on the occurrence of restenosis in femoro-popliteal (FP) arteries. The atheroprone flow conditions that are generally accepted to promote restenosis can be calculated by computational fluid dynamics (CFD) analyses, and these results can be used to assess individualized treatment outcomes. However, the impact of endovascular therapy on the flow behaviors of FP arteries are still poorly understood, as the imaging modalities used in existing numerical works (X-ray angiography, computed tomography angiography) are unable to accurately represent the post-treatment arterial geometry due to their low resolutions. Therefore, this study proposes a new algorithm that combines intra-arterial lumen geometry obtained from high-resolution optical coherence tomography (OCT) images with centerlines generated from X-ray images to reconstruct the FP artery with an in-plane resolution of 10 µm. This superior accuracy allows modeling characteristic geometrical structures, such as angioplasty-induced arterial dissections, that are too small to be reconstructed with other imaging modalities. The framework is applied on the clinical data of patients treated either with only-percutaneous transluminal angioplasty (PTA) (n = 4) or PTA followed by stenting (n = 4). Based on the generated models, PTA was found to cause numerous arterial dissections, covering approximately 10% of the total surface area of the lumen, whereas no dissections were identified in the stented arteries. CFD simulations were performed to investigate the hemodynamic conditions before and after treatment. Regardless of the treatment method, the areas affected by low time-averaged wall shear stress (< 0.5 Pa) were significantly higher (p < 0.05) following endovascular therapy (pre-PTA: 0.95 ± 0.59 cm2; post-PTA: 2.10 ± 1.09cm2; post-stent: 3.10 ± 0.98 cm2). There were no statistical differences between the PTA and the stent groups. However, within the PTA group, adverse hemodynamics were mainly concentrated at regions created by arterial dissections, which may negatively impact the outcomes of a leave-nothing-behind strategy. These observations show that OCT-based numerical models have great potential to guide clinicians regarding the optimal treatment approach.

Coaxial micro-extrusion of a calcium phosphate ink with aqueous solvents improves printing stability, structure fidelity and mechanical properties.

Micro-extrusion-based 3D printing of complex geometrical and porous calcium phosphate (CaP) can improve treatment of bone defects through the production of personalized bone substitutes. However, achieving printing and post-printing shape stabilities for the efficient fabrication and application of rapid hardening protocol are still challenging. In this work, the coaxial printing of a self-setting CaP cement with water and ethanol mixtures aiming to increase the ink yield stress upon extrusion and the stability of fabricated structures was explored. Printing height of overhang structure was doubled when aqueous solvents were used and a 2 log increase of the stiffness was achieved post-printing. A standard and fast steam sterilization protocol applied as hardening step on the coaxial printed CaP cement (CPC) ink resulted in constructs with 4 to 5 times higher compressive moduli in comparison to extrusion process in the absence of solvent. This improved mechanical performance is likely due to rapid CPC setting, preventing cracks formation during hardening process. Thus, coaxial micro-extrusion-based 3D printing of a CPC ink with aqueous solvent enhances printability and allows the use of the widespread steam sterilization cycle as a standalone post-processing technique for production of 3D printed personalized CaP bone substitutes. STATEMENT OF SIGNIFICANCE: Coaxial micro-extrusion-based 3D printing of a self-setting CaP cement with water:ethanol mixtures increased the ink yield stress upon extrusion and the stability of fabricated structures. Printing height of overhang structure was doubled when aqueous solvents were used, and a 2 orders of magnitude log increase of the stiffness was achieved post-printing. A fast hardening step consisting of a standard steam sterilization was applied. Four to 5 times higher compressive moduli was obtained for hardened coaxially printed constructs. This improved mechanical performance is likely due to rapid CPC setting in the coaxial printing, preventing cracks formation during hardening process.

Oxygen Kinetics During Corneal Cross-linking With and Without Supplementary Oxygen.

PURPOSE: To measure and simulate oxygen kinetics during corneal cross-linking at different irradiances with and without supplementary oxygen. DESIGN: Experimental, laboratory study. METHODS: In de-epithelialized porcine eyes, a femtosecond-laser-generated tunnel was used to place a fiber probe in corneal depths of 100, 200, and 300 µm to measure the local oxygen concentration. After riboflavin imbibition, the corneas were irradiated at 3, 9, 18, and 30 mW/cm2 while the oxygen concentration was measured. All experiments were performed under normoxic (21%) and hyperoxic (>95%) conditions. The obtained data were used to identify parameters of a numerical model for oxygen consumption and diffusion. RESULTS: The equilibrium stromal oxygen concentration under atmospheric oxygen at 3 mW/cm2 was 2.3% in 100 µm decreasing to <1% in 300 µm. With 9, 18, and 30 mW/cm2, no oxygen was available in 200 µm, respectively, 100 µm or deeper. Using a hyperoxic environment, the concentration was 50% using 3 mW/cm2 in 100 µm, decreasing to 40% in 300 µm. At 9 mW/cm2, the concentrations were 5%, 3%, and 1% in 100, 200 and 300 µm, respectively. Using 18 and 30 mW/cm2, all oxygen was depleted at 100 µm; however, oxygen half-lives were longer at 18 mW/cm2 than at 30 mW/cm2. The oxygen model was able to reproduce the experiments and indicated an exponential decay with increasing distance to the anterior surface. CONCLUSION: Supplementary oxygen increases the oxygen availability during corneal cross-linking. At higher irradiances, supplementary oxygen is beneficial and eliminates the bottleneck of oxygen allowing a potentially more efficient cross-linking. The calibrated numerical model can quantify the spatial oxygen concentration related to different scenarios such as irradiance or environmental oxygen concentration.

Deep learning for the rapid automatic quantification and characterization of rotator cuff muscle degeneration from shoulder CT datasets.

OBJECTIVES: This study aimed at developing a convolutional neural network (CNN) able to automatically quantify and characterize the level of degeneration of rotator cuff (RC) muscles from shoulder CT images including muscle atrophy and fatty infiltration. METHODS: One hundred three shoulder CT scans from 95 patients with primary glenohumeral osteoarthritis undergoing anatomical total shoulder arthroplasty were retrospectively retrieved. Three independent radiologists manually segmented the premorbid boundaries of all four RC muscles on standardized sagittal-oblique CT sections. This premorbid muscle segmentation was further automatically predicted using a CNN. Automatically predicted premorbid segmentations were then used to quantify the ratio of muscle atrophy, fatty infiltration, secondary bone formation, and overall muscle degeneration. These muscle parameters were compared with measures obtained manually by human raters. RESULTS: Average Dice similarity coefficients for muscle segmentations obtained automatically with the CNN (88% ± 9%) and manually by human raters (89% ± 6%) were comparable. No significant differences were observed for the subscapularis, supraspinatus, and teres minor muscles (p > 0.120), whereas Dice coefficients of the automatic segmentation were significantly higher for the infraspinatus (p < 0.012). The automatic approach was able to provide good-very good estimates of muscle atrophy (R2 = 0.87), fatty infiltration (R2 = 0.91), and overall muscle degeneration (R2 = 0.91). However, CNN-derived segmentations showed a higher variability in quantifying secondary bone formation (R2 = 0.61) than human raters (R2 = 0.87). CONCLUSIONS: Deep learning provides a rapid and reliable automatic quantification of RC muscle atrophy, fatty infiltration, and overall muscle degeneration directly from preoperative shoulder CT scans of osteoarthritic patients, with an accuracy comparable with that of human raters. KEY POINTS: • Deep learning can not only segment RC muscles currently available in CT images but also learn their pre-existing locations and shapes from invariant anatomical structures visible on CT sections. • Our automatic method is able to provide a rapid and reliable quantification of RC muscle atrophy and fatty infiltration from conventional shoulder CT scans. • The accuracy of our automatic quantitative technique is comparable with that of human raters.

Corneal Biomechanics After Intrastromal Ring Surgery: Optomechanical In Silico Assessment.

Purpose: To provide a biomechanical framework to better understand the postsurgical optomechanical behavior of the cornea after ring implantation. Methods: Calibrated in silico models were used to determine the corneal shape and stresses after ring implantation. After mechanical simulations, geometric ray-tracing was used to determine the change in spherical equivalent. The effect of the surgical procedure, circadian variation of intraocular pressure, or the biomechanical weakening introduced by keratoconus (KC) were evaluated for each intrastromal ring. Results: Models predicted the postsurgical optomechanical response of the cornea at a population level. The localized mechanical effect of the additional intrastromal volume introduced by the implants (size and diameter) drives the postsurgical corneal response. However, central corneal stresses did not increase more than 50%, and thus implants did not strengthen the cornea globally. Because of the biomechanical weakening introduced by laser pocketing, continuous implants in a pocket resulted in higher refractive corrections and in the relaxation of the anterior stroma, which could slow down KC progression. Implants can move within the stroma, acting as a dynamic pivot point that modifies corneal kinematics and flattens the corneal center. Changes in stromal mechanical properties did not impact on refraction for normal or pathological corneas. Conclusions: Implants do not stiffen the cornea but create a local bulkening effect that regularizes the corneal shape by modifying corneal kinematics without canceling corneal motion. Translational Relevance: In silico models can help to understand corneal biomechanics, to plan patient-specific interventions, or to create biomechanically driven nomograms.

Automated CT bone segmentation using statistical shape modelling and local template matching.

Accurate CT bone segmentation is essential to develop chair-side manufacturing of implants based on additive manufacturing. We herewith present an automated method able to accurately segment challenging bone regions, while simultaneously providing anatomical correspondences. The method was evaluated on demanding regions: normal and osteoarthritic scapulae, healthy and atrophied mandibles, and orbital bones. On average, results were accurate with surface distances of approximately 0.5 mm and average Dice coefficients >90%. Since anatomical correspondences are propagated during the segmentation process, this approach can directly yield anatomical measurements, provide design parameters for personalized surgical instruments, or determine the bone geometry to manufacture patient-specific implants.

Optimization of surgical parameters based on patient-specific models: Application to arcuate keratotomy.

PURPOSE: To determine surgical parameters for arcuate keratotomy by simulating the intervention with a patient-specific model. SETTING: University Eye Clinic Salzburg, Paracelsus Medical University, Austria, and Institute for Surgical Technology and Biomechanics, University of Bern, Switzerland. DESIGN: Computational modeling study. METHODS: A new approach to plan arcuate keratotomy based on personalized finite element simulations was developed. Using this numeric tool, an optimization algorithm was implemented to determine the incision parameters that best met the surgeon's requirements while preserving the orientation of the astigmatism. Virtual surgeries were performed on patients to compare the performance of the simulation-based approach with results based on the Lindstrom and Donnenfeld nomograms and with intrastromal interventions. RESULTS: Retrospective data on 28 patients showed that personalized simulation reproduced the surgically induced change in astigmatism (Pearson correlation = 0.8). Patient-specific simulation was used to examine strategies for arcuate interventions on 621 corneal topographies. The Lindstrom nomogram resulted in low postoperative astigmatism (mean 0.03 diopter [D] ± 0.3 [SD]) but frequent overcorrections (20%). The Donnenfeld nomogram and intrastromal incisions resulted in a small amount of overcorrection (1.5%) but a wider spread in astigmatism (mean 0.63 ± 0.35 D and 0.48 ± 0.50 D, respectively). In contrast, the new numeric parameter optimization approach led to postoperative astigmatism values (mean 0.40 ± 0.08 D, 0.20 ± 0.08 D, and 0.04 ± 0.13 D) that closely matched the target astigmatism (0.40 D, 0.20 D, and 0.00 D), respectively, while keeping the number of overcorrections low (<1.5%). CONCLUSION: Using numeric modeling to optimize surgical parameters for arcuate keratotomy led to more reliable postoperative astigmatism, limiting the risk for overcorrection.

Prediction of restenosis based on hemodynamical markers in revascularized femoro-popliteal arteries during leg flexion.

Endovascular therapy in patients suffering from peripheral arterial disease shows high rates of restenosis. The poor clinical outcomes are commonly explained by the demanding mechanical environment due to leg movements, but the mechanisms responsible for restenosis remain unknown. In this study, we hypothesized that restenosis following revascularization is associated with hemodynamical markers derived from blood flow during leg flexion. Therefore, we performed personalized computational fluid dynamics (CFD) analyses of 20 patients, who underwent routine endovascular femoro-popliteal interventions. The CFD analyses were conducted using 3D models of the arterial geometry in straight and flexed positions, which were reconstructed from 2D angiographic images. Based on restenosis rates reported at 6-month follow-up, logistic regression analyses were performed to predict restenosis from hemodynamical parameters. Results showed that severe arterial deformations, such as kinking, induced by leg flexion in stented arteries led to adverse hemodynamic conditions that may trigger restenosis. A logistic regression analysis based solely on hemodynamical markers had an accuracy of 75%, which showed that flow parameters are sufficient to predict restenosis (p = 0.031). However, better predictions were achieved by adding the treatment method in the model. This suggests that a more accurate image acquisition technique is required to capture the localized modifications of blood flow following intervention, especially around the stented artery. This approach, based on the immediate postoperative configuration of the artery, has the potential to identify patients at increased risk of restenosis. Based on this information, clinicians could take preventive measures and more closely follow these patients to avoid complications.