MSO-GP: 3-D segmentation of large and complex conjoined tree structures.

Robust segmentation of large and complex conjoined tree structures in 3-D is a major challenge in computer vision. This is particularly true in computational biology, where we often encounter large data structures in size, but few in number, which poses a hard problem for learning algorithms. We show that merging multiscale opening with geodesic path propagation, can shed new light on this classic machine vision challenge, while circumventing the learning issue by developing an unsupervised visual geometry approach (digital topology/morphometry). The novelty of the proposed MSO-GP method comes from the geodesic path propagation being guided by a skeletonization of the conjoined structure that helps to achieve robust segmentation results in a particularly challenging task in this area, that of artery-vein separation from non-contrast pulmonary computed tomography angiograms. This is an important first step in measuring vascular geometry to then diagnose pulmonary diseases and to develop image-based phenotypes. We first present proof-of-concept results on synthetic data, and then verify the performance on pig lung and human lung data with less segmentation time and user intervention needs than those of the competing methods.

Artificial Intelligence Applied to Osteoporosis: A Performance Comparison of Machine Learning Algorithms in Predicting Fragility Fractures From MRI Data.

BACKGROUND: A current challenge in osteoporosis is identifying patients at risk of bone fracture. PURPOSE: To identify the machine learning classifiers that predict best osteoporotic bone fractures and, from the data, to highlight the imaging features and the anatomical regions that contribute most to prediction performance. STUDY TYPE: Prospective (cross-sectional) case-control study. POPULATION: Thirty-two women with prior fragility bone fractures, of mean age = 61.6 and body mass index (BMI) = 22.7 kg/m2 , and 60 women without fractures, of mean age = 62.3 and BMI = 21.4 kg/m2 . Field Strength/ Sequence: 3D FLASH at 3T. ASSESSMENT: Quantitative MRI outcomes by software algorithms. Mechanical and topological microstructural parameters of the trabecular bone were calculated for five femoral regions, and added to the vector of features together with bone mineral density measurement, fracture risk assessment tool (FRAX) score, and personal characteristics such as age, weight, and height. We fitted 15 classifiers using 200 randomized cross-validation datasets. Statistical Tests: Data: Kolmogorov-Smirnov test for normality. Model Performance: sensitivity, specificity, precision, accuracy, F1-test, receiver operating characteristic curve (ROC). Two-sided t-test, with P < 0.05 for statistical significance. RESULTS: The top three performing classifiers are RUS-boosted trees (in particular, performing best with head data, F1 = 0.64 ± 0.03), the logistic regression and the linear discriminant (both best with trochanteric datasets, F1 = 0.65 ± 0.03 and F1 = 0.67 ± 0.03, respectively). A permutation of these classifiers comprised the best three performers for four out of five anatomical datasets. After averaging across all the anatomical datasets, the score for the best performer, the boosted trees, was F1 = 0.63 ± 0.03 for All-features dataset, F1 = 0.52 ± 0.05 for the no-MRI dataset, and F1 = 0.48 ± 0.06 for the no-FRAX dataset. Data Conclusion: Of many classifiers, the RUS-boosted trees, the logistic regression, and the linear discriminant are best for predicting osteoporotic fracture. Both MRI and FRAX independently add value in identifying osteoporotic fractures. The femoral head, greater trochanter, and inter-trochanter anatomical regions within the proximal femur yielded better F1-scores for the best three classifiers. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:1029-1038.

3-T MR Imaging of Proximal Femur Microarchitecture in Subjects with and without Fragility Fracture and Nonosteoporotic Proximal Femur Bone Mineral Density.

Purpose To determine if 3-T magnetic resonance (MR) imaging of proximal femur microarchitecture can allow discrimination of subjects with and without fragility fracture who do not have osteoporotic proximal femur bone mineral density (BMD). Materials and Methods Sixty postmenopausal women (30 with and 30 without fragility fracture) who had BMD T scores of greater than -2.5 in the hip were recruited. All subjects underwent dual-energy x-ray absorptiometry to assess BMD and 3-T MR imaging of the same hip to assess bone microarchitecture. World Health Organization Fracture Risk Assessment Tool (FRAX) scores were also computed. We used the Mann-Whitney test, receiver operating characteristics analyses, and Spearman correlation estimates to assess differences between groups, discriminatory ability with parameters, and correlations among BMD, microarchitecture, and FRAX scores. Results Patients with versus without fracture showed a lower trabecular plate-to-rod ratio (median, 2.41 vs 4.53, respectively), lower trabecular plate width (0.556 mm vs 0.630 mm, respectively), and lower trabecular thickness (0.114 mm vs 0.126 mm) within the femoral neck, and higher trabecular rod disruption (43.5 vs 19.0, respectively), higher trabecular separation (0.378 mm vs 0.323 mm, respectively), and lower trabecular number (0.158 vs 0.192, respectively), lower trabecular connectivity (0.015 vs 0.027, respectively) and lower trabecular plate-to-rod ratio (6.38 vs 8.09, respectively) in the greater trochanter (P < .05 for all). Trabecular plate-to-rod ratio, plate width, and thickness within the femoral neck (areas under the curve [AUCs], 0.654-0.683) and trabecular rod disruption, number, connectivity, plate-to-rod ratio, and separation within the greater trochanter (AUCs, 0.662-0.694) allowed discrimination of patients with fracture from control subjects. Femoral neck, total hip, and spine BMD did not differ between and did not allow discrimination between groups. FRAX scores including and not including BMD allowed discrimination between groups (AUCs, 0.681-0.773). Two-factor models (one MR imaging microarchitectural parameter plus a FRAX score without BMD) allowed discrimination between groups (AUCs, 0.702-0.806). There were no linear correlations between BMD and microarchitectural parameters (Spearman ρ, -0.198 to 0.196). Conclusion 3-T MR imaging of proximal femur microarchitecture allows discrimination between subjects with and without fragility fracture who have BMD T scores of greater than -2.5 and may provide different information about bone quality than that provided by dual-energy x-ray absorptiometry. © RSNA, 2018.

Subchondral bone microarchitecture analysis in the proximal tibia at 7-T MRI.

Background Bone remodels in response to mechanical loads and osteoporosis results from impaired ability of bone to remodel. Bone microarchitecture analysis provides information on bone quality beyond bone mineral density (BMD). Purpose To compare subchondral bone microarchitecture parameters in the medial and lateral tibia plateau in individuals with and without fragility fractures. Material and Methods Twelve female patients (mean age = 58 ± 15 years; six with and six without previous fragility fractures) were examined with dual-energy X-ray absorptiometry (DXA) and 7-T magnetic resonance imaging (MRI) of the proximal tibia. A transverse high-resolution three-dimensional fast low-angle shot sequence was acquired (0.234 × 0.234 × 1 mm). Digital topological analysis (DTA) was applied to the medial and lateral subchondral bone of the proximal tibia. The following DTA-based bone microarchitecture parameters were assessed: apparent bone volume; trabecular thickness; profile-edge-density (trabecular bone erosion parameter); profile-interior-density (intact trabecular rods parameter); plate-to-rod ratio; and erosion index. We compared femoral neck T-scores and bone microarchitecture parameters between patients with and without fragility fracture. Results There was no statistical significant difference in femoral neck T-scores between individuals with and without fracture (-2.4 ± 0.9 vs. -1.8 ± 0.7, P = 0.282). Apparent bone volume in the medial compartment was lower in patients with previous fragility fracture (0.295 ± 0.022 vs. 0.317 ± 0.009; P = 0.016). Profile-edge-density, a trabecular bone erosion parameter, was higher in patients with previous fragility fracture in the medial (0.008 ± 0.003 vs. 0.005 ± 0.001) and lateral compartment (0.008 ± 0.002 vs. 0.005 ± 0.001); both P = 0.025. Other DTA parameters did not differ between groups. Conclusion 7-T MRI and DTA permit detection of subtle changes in subchondral bone quality when differences in BMD are not evident.

7T MRI of distal radius trabecular bone microarchitecture: How trabecular bone quality varies depending on distance from end-of-bone.

PURPOSE: To use 7T magnetic resonance imaging (MRI) to determine how trabecular bone microarchitecture varies at the epiphysis, metaphysis, and diaphysis of the distal radius. MATERIALS AND METHODS: The distal radius of 24 females (mean age = 56 years, range = 24-78 years) was scanned on a 7T MRI using a 3D fast low-angle shot sequence (0.169 × 0.169 × 1 mm). Digital topological analysis was applied at the epiphysis, metaphysis, and diaphysis to compute: total trabecular bone volume; trabecular thickness, number, connectivity, and erosion index (a measure of network resorption). Differences and correlations were assessed using standard statistical methods. RESULTS: The metaphysis and epiphysis had 83-123% greater total bone volume and 14-16% greater trabecular number than the diaphysis (both P < 0.0001). The erosion index was significantly higher at the diaphysis than the metaphysis and epiphysis (both P < 0.01). The most elderly volunteers had lower trabecular number (<66 years mean 0.29 ± 0.01; ≥66 years, 0.27 ± 0.02, P < 0.05) and higher erosion index (<66 years mean 1.18 ± 0.17; age ≥66 years, mean 1.42 ± 0.46, P < 0.05) at the epiphysis; differences not detected by total trabecular bone volume. CONCLUSION: 7T MRI reveals trabecular bone microarchitecture varies depending on scan location at the end-of-bone, being of overall higher quality distally (epiphysis) than proximally (diaphysis). Age-related differences in trabecular microarchitecture can be detected by 7T MRI. The results highlight the potential sensitivity of 7T MRI to microarchitectural differences and the potential importance of standardizing scan location for future clinical studies of fracture risk or treatment response. LEVEL OF EVIDENCE: 3 J. Magn. Reson. Imaging 2017;45:872-878.

Quantitative Dual-Energy Computed Tomography Supports a Vascular Etiology of Smoking-induced Inflammatory Lung Disease.

RATIONALE: Endothelial dysfunction is of interest in relation to smoking-associated emphysema, a component of chronic obstructive pulmonary disease (COPD). We previously demonstrated that computed tomography (CT)-derived pulmonary blood flow (PBF) heterogeneity is greater in smokers with normal pulmonary function tests (PFTs) but who have visual evidence of centriacinar emphysema (CAE) on CT. OBJECTIVES: We introduced dual-energy CT (DECT) perfused blood volume (PBV) as a PBF surrogate to evaluate whether the CAE-associated increased PBF heterogeneity is reversible with sildenafil. METHODS: Seventeen PFT-normal current smokers were divided into CAE-susceptible (SS; n = 10) and nonsusceptible (NS; n = 7) smokers, based on the presence or absence of CT-detected CAE. DECT-PBV images were acquired before and 1 hour after administration of 20 mg oral sildenafil. Regional PBV and PBV coefficients of variation (CV), a measure of spatial blood flow heterogeneity, were determined, followed by quantitative assessment of the central arterial tree. MEASUREMENTS AND MAIN RESULTS: After sildenafil administration, regional PBV-CV decreased in SS subjects but did not decrease in NS subjects (P < 0.05), after adjusting for age and pack-years. Quantitative evaluation of the central pulmonary arteries revealed higher arterial volume and greater cross-sectional area (CSA) in the lower lobes of SS smokers, which suggested arterial enlargement in response to increased peripheral resistance. After sildenafil, arterial CSA decreased in SS smokers but did not decrease in NS smokers (P < 0.01). CONCLUSIONS: These results demonstrate that sildenafil restores peripheral perfusion and reduces central arterial enlargement in normal SS subjects with little effect in NS subjects, highlighting DECT-PBV as a biomarker of reversible endothelial dysfunction in smokers with CAE.

Multiscale Opening of Conjoined Fuzzy Objects: Theory and Applications.

Theoretical properties of a multi-scale opening (MSO) algorithm for two conjoined fuzzy objects are established, and its extension to separating two conjoined fuzzy objects with different intensity properties is introduced. Also, its applications to artery/vein (A/V) separation in pulmonary CT imaging and carotid vessel segmentation in CT angiograms (CTAs) of patients with intracranial aneurysms are presented. The new algorithm accounts for distinct intensity properties of individual conjoined objects by combining fuzzy distance transform (FDT), a morphologic feature, with fuzzy connectivity, a topologic feature. The algorithm iteratively opens the two conjoined objects starting at large scales and progressing toward finer scales. Results of application of the method in separating arteries and veins in a physical cast phantom of a pig lung are presented. Accuracy of the algorithm is quantitatively evaluated in terms of sensitivity and specificity on patients' CTA data sets and its performance is compared with existing methods. Reproducibility of the algorithm is examined in terms of volumetric agreement between two users' carotid vessel segmentation results. Experimental results using this algorithm on patients' CTA data demonstrate a high average accuracy of 96.3% with 95.1% sensitivity and 97.5% specificity and a high reproducibility of 94.2% average agreement between segmentation results from two mutually independent users. Approximately, twenty-five to thirty-five user-specified seeds/separators are needed for each CTA data through a custom designed graphical interface requiring an average of thirty minutes to complete carotid vascular segmentation in a patient's CTA data set.

A Robust and Efficient Curve Skeletonization Algorithm for Tree-Like Objects Using Minimum Cost Paths.

Conventional curve skeletonization algorithms using the principle of Blum's transform, often, produce unwanted spurious branches due to boundary irregularities, digital effects, and other artifacts. This paper presents a new robust and efficient curve skeletonization algorithm for three-dimensional (3-D) elongated fuzzy objects using a minimum cost path approach, which avoids spurious branches without requiring post-pruning. Starting from a root voxel, the method iteratively expands the skeleton by adding new branches in each iteration that connects the farthest quench voxel to the current skeleton using a minimum cost path. The path-cost function is formulated using a novel measure of local significance factor defined by the fuzzy distance transform field, which forces the path to stick to the centerline of an object. The algorithm terminates when dilated skeletal branches fill the entire object volume or the current farthest quench voxel fails to generate a meaningful skeletal branch. Accuracy of the algorithm has been evaluated using computer-generated phantoms with known skeletons. Performance of the method in terms of false and missing skeletal branches, as defined by human experts, has been examined using in vivo CT imaging of human intrathoracic airways. Results from both experiments have established the superiority of the new method as compared to the existing methods in terms of accuracy as well as robustness in detecting true and false skeletal branches. The new algorithm makes a significant reduction in computation complexity by enabling detection of multiple new skeletal branches in one iteration. Specifically, this algorithm reduces the number of iterations from the number of terminal tree branches to the worst case performance of tree depth. In fact, experimental results suggest that, on an average, the order of computation complexity is reduced to the logarithm of the number of terminal branches of a tree-like object.

In vivo measurement reproducibility of femoral neck microarchitectural parameters derived from 3T MR images.

PURPOSE: To evaluate the within-day and between-day measurement reproducibility of in vivo 3D MRI assessment of trabecular bone microarchitecture of the proximal femur. MATERIALS AND METHODS: This Health Insurance Portability and Accountability Act (HIPPA)-compliant, Institutional Review Board (IRB)-approved study was conducted on 11 healthy subjects (mean age = 57.4 ± 14.1 years) with written informed consent. All subjects underwent a 3T MRI hip scan in vivo (0.234 × 0.234 × 1.5 mm) at three timepoints: baseline, second scan same day (intrascan), and third scan 1 week later (interscan). We applied digital topological analysis and volumetric topological analysis to compute the following microarchitectural parameters within the femoral neck: total bone volume, bone volume fraction, markers of trabecular number (skeleton density), connectivity (junctions), plate-like structure (surfaces), plate width, and trabecular thickness. Reproducibility was assessed using root-mean-square coefficient of variation (RMS-CV) and intraclass correlation coefficient (ICC). RESULTS: The within-day RMS-CVs ranged from 2.3% to 7.8%, and the between-day RMS-CVs ranged from 4.0% to 7.3% across all parameters. The within-day ICCs ranged from 0.931 to 0.989, and the between-day ICCs ranged from 0.934 to 0.971 across all parameters. CONCLUSION: These results demonstrate high reproducibility for trabecular bone microarchitecture measures derived from 3T MR images of the proximal femur. The measurement reproducibility is within a range suitable for clinical cross-sectional and longitudinal studies in osteoporosis.

7T MRI detects deterioration in subchondral bone microarchitecture in subjects with mild knee osteoarthritis as compared with healthy controls.

PURPOSE: To determine how subchondral bone microarchitecture is altered in patients with mild knee osteoarthritis. MATERIALS AND METHODS: This study had Institutional Review Board approval. We recruited 24 subjects with mild radiographic knee osteoarthritis and 16 healthy controls. The distal femur was scanned at 7T using a high-resolution 3D FLASH sequence. We applied digital topological analysis to assess bone volume fraction, markers of trabecular number (skeleton density), trabecular network osteoclastic resorption (erosion index), plate-like structure (surface), rod-like structure (curve), and plate-to-rod ratio (surface-curve ratio). We used two-tailed t-tests to compare differences between osteoarthritis subjects and controls. RESULTS: 7T magnetic resonance imaging (MRI) detected deterioration in subchondral bone microarchitecture in both medial and lateral femoral condyles in osteoarthritis subjects as compared with controls. This was manifested by lower bone volume fraction (-1.03% to -5.43%, P < 0.04), higher erosion index (+8.49 to +22.76%, P < 0.04), lower surface number (-2.31% to -9.63%, P < 0.007), higher curve number (+6.85% to +16.93%, P < 0.03), and lower plate-to-rod ratio (-7.92% to -21.71%, P < 0.05). CONCLUSION: The results provide further support for the concept that poor subchondral bone quality is associated with osteoarthritis and may serve as a potential therapeutic target for osteoarthritis interventions. J. Magn. Reson. Imaging 2015;41:1311-1317. © 2014 Wiley Periodicals, Inc.

7 Tesla MRI of bone microarchitecture discriminates between women without and with fragility fractures who do not differ by bone mineral density.

Osteoporosis is a disease of poor bone quality. Bone mineral density (BMD) has limited ability to discriminate between subjects without and with poor bone quality, and assessment of bone microarchitecture may have added value in this regard. Our goals were to use 7 T MRI to: (1) quantify and compare distal femur bone microarchitecture in women without and with poor bone quality (defined clinically by presence of fragility fractures); and (2) determine whether microarchitectural parameters could be used to discriminate between these two groups. This study had institutional review board approval, and we obtained written informed consent from all subjects. We used a 28-channel knee coil to image the distal femur of 31 subjects with fragility fractures and 25 controls without fracture on a 7 T MRI scanner using a 3-D fast low angle shot sequence (0.234 mm × 0.234 mm × 1 mm, parallel imaging factor = 2, acquisition time = 7 min 9 s). We applied digital topological analysis to quantify parameters of bone microarchitecture. All subjects also underwent standard clinical BMD assessment in the hip and spine. Compared to controls, fracture cases demonstrated lower bone volume fraction and markers of trabecular number, plate-like structure, and plate-to-rod ratio, and higher markers of trabecular isolation, rod disruption, and network resorption (p < 0.05 for all). There were no differences in hip or spine BMD T-scores between groups (p > 0.05). In receiver-operating-characteristics analyses, microarchitectural parameters could discriminate cases and controls (AUC = 0.66-0.73, p < 0.05). Hip and spine BMD T-scores could not discriminate cases and controls (AUC = 0.58-0.64, p ≥ 0.08). We conclude that 7 T MRI can detect bone microarchitectural deterioration in women with fragility fractures who do not differ by BMD. Microarchitectural parameters might some day be used as an additional tool to detect patients with poor bone quality who cannot be detected by dual-energy X-ray absorptiometry (DXA).

Assessment of morphometry of pulmonary acini in mouse lungs by nondestructive imaging using multiscale microcomputed tomography.

Establishing the 3D architecture and morphometry of the intact pulmonary acinus is an essential step toward a more complete understanding of the relationship of lung structure and function. We combined a special fixation method with a unique volumetric nondestructive imaging technique and image processing tools to separate individual acini in the mouse lung. Interior scans of the parenchyma at a resolution of 2 µm enabled the reconstruction and quantitative study of whole acini by image analysis and stereologic methods, yielding data characterizing the 3D morphometry of the pulmonary acinus. The 3D reconstructions compared well with the architecture of silicon rubber casts of mouse acini. The image-based segmentation of individual acini allowed the computation of acinar volume and surface area, as well as estimation of the number of alveoli per acinus using stereologic methods. The acinar morphometry of male C57BL/6 mice age 12 wk and 91 wk was compared. Significant increases in all parameters as a function of age suggest a continuous change of the lung morphometry, with an increase in alveoli beyond what has been previously viewed as the maturation phase of the animals. Our image analysis methods open up opportunities for defining and quantitatively assessing the acinar structure in healthy and diseased lungs. The methods applied here to mice can be adjusted for the study of similarly prepared human lungs.

Chest CT-based automated vertebral fracture assessment using artificial intelligence and morphologic features.

BACKGROUND: Spinal degeneration and vertebral compression fractures are common among the elderly that adversely affect their mobility, quality of life, lung function, and mortality. Assessment of vertebral fractures in chronic obstructive pulmonary disease (COPD) is important due to the high prevalence of osteoporosis and associated vertebral fractures in COPD. PURPOSE: We present new automated methods for (1) segmentation and labelling of individual vertebrae in chest computed tomography (CT) images using deep learning (DL), multi-parametric freeze-and-grow (FG) algorithm, and separation of apparently fused vertebrae using intensity autocorrelation and (2) vertebral deformity fracture detection using computed vertebral height features and parametric computational modelling of an established protocol outlined for trained human experts. METHODS: A chest CT-based automated method was developed for quantitative deformity fracture assessment following the protocol by Genant et al. The computational method was accomplished in the following steps: (1) computation of a voxel-level vertebral body likelihood map from chest CT using a trained DL network; (2) delineation and labelling of individual vertebrae on the likelihood map using an iterative multi-parametric FG algorithm; (3) separation of apparently fused vertebrae in CT using intensity autocorrelation; (4) computation of vertebral heights using contour analysis on the central anterior-posterior (AP) plane of a vertebral body; (5) assessment of vertebral fracture status using ratio functions of vertebral heights and optimized thresholds. The method was applied to inspiratory or total lung capacity (TLC) chest scans from the multi-site Genetic Epidemiology of COPD (COPDGene) (ClinicalTrials.gov: NCT00608764) study, and the performance was examined (n = 3231). One hundred and twenty scans randomly selected from this dataset were partitioned into training (n = 80) and validation (n = 40) datasets for the DL-based vertebral body classifier. Also, generalizability of the method to low dose CT imaging (n = 236) was evaluated. RESULTS: The vertebral segmentation module achieved a Dice score of .984 as compared to manual outlining results as reference (n = 100); the segmentation performance was consistent across images with the minimum and maximum of Dice scores among images being .980 and .989, respectively. The vertebral labelling module achieved 100% accuracy (n = 100). For low dose CT, the segmentation module produced image-level minimum and maximum Dice scores of .995 and .999, respectively, as compared to standard dose CT as the reference; vertebral labelling at low dose CT was fully consistent with standard dose CT (n = 236). The fracture assessment method achieved overall accuracy, sensitivity, and specificity of 98.3%, 94.8%, and 98.5%, respectively, for 40,050 vertebrae from 3231 COPDGene participants. For generalizability experiments, fracture assessment from low dose CT was consistent with the reference standard dose CT results across all participants. CONCLUSIONS: Our CT-based automated method for vertebral fracture assessment is accurate, and it offers a feasible alternative to manual expert reading, especially for large population-based studies, where automation is important for high efficiency. Generalizability of the method to low dose CT imaging further extends the scope of application of the method, particularly since the usage of low dose CT imaging in large population-based studies has increased to reduce cumulative radiation exposure.

Deep learning-based harmonization of trabecular bone microstructures between high- and low-resolution CT imaging.

BACKGROUND: Osteoporosis is a bone disease related to increased bone loss and fracture-risk. The variability in bone strength is partially explained by bone mineral density (BMD), and the remainder is contributed by bone microstructure. Recently, clinical CT has emerged as a viable option for in vivo bone microstructural imaging. Wide variations in spatial-resolution and other imaging features among different CT scanners add inconsistency to derived bone microstructural metrics, urging the need for harmonization of image data from different scanners. PURPOSE: This paper presents a new deep learning (DL) method for the harmonization of bone microstructural images derived from low- and high-resolution CT scanners and evaluates the method's performance at the levels of image data as well as derived microstructural metrics. METHODS: We generalized a three-dimensional (3D) version of GAN-CIRCLE that applies two generative adversarial networks (GANs) constrained by the identical, residual, and cycle learning ensemble (CIRCLE). Two GAN modules simultaneously learn to map low-resolution CT (LRCT) to high-resolution CT (HRCT) and vice versa. Twenty volunteers were recruited. LRCT and HRCT scans of the distal tibia of their left legs were acquired. Five-hundred pairs of LRCT and HRCT image blocks of 64 × 64 × 64 $64 \times 64 \times 64 $ voxels were sampled for each of the twelve volunteers and used for training in supervised as well as unsupervised setups. LRCT and HRCT images of the remaining eight volunteers were used for evaluation. LRCT blocks were sampled at 32 voxel intervals in each coordinate direction and predicted HRCT blocks were stitched to generate a predicted HRCT image. RESULTS: Mean ± standard deviation of structural similarity (SSIM) values between predicted and true HRCT using both 3DGAN-CIRCLE-based supervised (0.84 ± 0.03) and unsupervised (0.83 ± 0.04) methods were significantly (p < 0.001) higher than the mean SSIM value between LRCT and true HRCT (0.75 ± 0.03). All Tb measures derived from predicted HRCT by the supervised 3DGAN-CIRCLE showed higher agreement (CCC  ∈ $ \in $ [0.956 0.991]) with the reference values from true HRCT as compared to LRCT-derived values (CCC  ∈ $ \in $ [0.732 0.989]). For all Tb measures, except Tb plate-width (CCC = 0.866), the unsupervised 3DGAN-CIRCLE showed high agreement (CCC  ∈ $ \in $ [0.920 0.964]) with the true HRCT-derived reference measures. Moreover, Bland-Altman plots showed that supervised 3DGAN-CIRCLE predicted HRCT reduces bias and variability in residual values of different Tb measures as compared to LRCT and unsupervised 3DGAN-CIRCLE predicted HRCT. The supervised 3DGAN-CIRCLE method produced significantly improved performance (p < 0.001) for all Tb measures as compared to the two DL-based supervised methods available in the literature. CONCLUSIONS: 3DGAN-CIRCLE, trained in either unsupervised or supervised fashion, generates HRCT images with high structural similarity to the reference true HRCT images. The supervised 3DGAN-CIRCLE improves agreements of computed Tb microstructural measures with their reference values and outperforms the unsupervised 3DGAN-CIRCLE. 3DGAN-CIRCLE offers a viable DL solution to retrospectively improve image resolution, which may aid in data harmonization in multi-site longitudinal studies where scanner mismatch is unavoidable.

Ultra-low dose hip CT-based automated measurement of volumetric bone mineral density at proximal femoral subregions.

BACKGROUND: Forty to fifty percent of women and 13%-22% of men experience an osteoporosis-related fragility fracture in their lifetimes. After the age of 50 years, the risk of hip fracture doubles in every 10 years. x-Ray based DXA is currently clinically used to diagnose osteoporosis and predict fracture risk. However, it provides only 2-D representation of bone and is associated with other technical limitations. Thus, alternative methods are needed. PURPOSE: To develop and evaluate an ultra-low dose (ULD) hip CT-based automated method for assessment of volumetric bone mineral density (vBMD) at proximal femoral subregions. METHODS: An automated method was developed to segment the proximal femur in ULD hip CT images and delineate femoral subregions. The computational pipeline consists of deep learning (DL)-based computation of femur likelihood map followed by shape model-based femur segmentation and finite element analysis-based warping of a reference subregion labeling onto individual femur shapes. Finally, vBMD is computed over each subregion in the target image using a calibration phantom scan. A total of 100 participants (50 females) were recruited from the Genetic Epidemiology of COPD (COPDGene) study, and ULD hip CT imaging, equivalent to 18 days of background radiation received by U.S. residents, was performed on each participant. Additional hip CT imaging using a clinical protocol was performed on 12 participants and repeat ULD hip CT was acquired on another five participants. ULD CT images from 80 participants were used to train the DL network; ULD CT images of the remaining 20 participants as well as clinical and repeat ULD CT images were used to evaluate the accuracy, generalizability, and reproducibility of segmentation of femoral subregions. Finally, clinical CT and repeat ULD CT images were used to evaluate accuracy and reproducibility of ULD CT-based automated measurements of femoral vBMD. RESULTS: Dice scores of accuracy (n = 20), reproducibility (n = 5), and generalizability (n = 12) of ULD CT-based automated subregion segmentation were 0.990, 0.982, and 0.977, respectively, for the femoral head and 0.941, 0.970, and 0.960, respectively, for the femoral neck. ULD CT-based regional vBMD showed Pearson and concordance correlation coefficients of 0.994 and 0.977, respectively, and a root-mean-square coefficient of variation (RMSCV) (%) of 1.39% with the clinical CT-derived reference measure. After 3-digit approximation, each of Pearson and concordance correlation coefficients as well as intraclass correlation coefficient (ICC) between baseline and repeat scans were 0.996 with RMSCV of 0.72%. Results of ULD CT-based bone analysis on 100 participants (age (mean ± SD) 73.6 ± 6.6 years) show that males have significantly greater (p < 0.01) vBMD at the femoral head and trochanteric regions than females, while females have moderately greater vBMD (p = 0.05) at the medial half of the femoral neck than males. CONCLUSION: Deep learning, combined with shape model and finite element analysis, offers an accurate, reproducible, and generalizable algorithm for automated segmentation of the proximal femur and anatomic femoral subregions using ULD hip CT images. ULD CT-based regional measures of femoral vBMD are accurate and reproducible and demonstrate regional differences between males and females.

Vessel and Airway Characteristics in One-Year Computed Tomography-defined Rapid Emphysema Progression: SPIROMICS.

Rationale: Rates of emphysema progression vary in chronic obstructive pulmonary disease (COPD), and the relationships with vascular and airway pathophysiology remain unclear. Objectives: We sought to determine if indices of peripheral (segmental and beyond) pulmonary arterial dilation measured on computed tomography (CT) are associated with a 1-year index of emphysema (EI; percentage of voxels <-950 Hounsfield units) progression. Methods: Five hundred ninety-nine former and never-smokers (Global Initiative for Chronic Obstructive Lung Disease stages 0-3) were evaluated from the SPIROMICS (Subpopulations and Intermediate Outcome Measures in COPD Study) cohort: rapid emphysema progressors (RPs; n = 188, 1-year ΔEI > 1%), nonprogressors (n = 301, 1-year ΔEI ± 0.5%), and never-smokers (n = 110). Segmental pulmonary arterial cross-sectional areas were standardized to associated airway luminal areas (segmental pulmonary artery-to-airway ratio [PAARseg]). Full-inspiratory CT scan-derived total (arteries and veins) pulmonary vascular volume (TPVV) was compared with small vessel volume (radius smaller than 0.75 mm). Ratios of airway to lung volume (an index of dysanapsis and COPD risk) were compared with ratios of TPVV to lung volume. Results: Compared with nonprogressors, RPs exhibited significantly larger PAARseg (0.73 ± 0.29 vs. 0.67 ± 0.23; P = 0.001), lower ratios of TPVV to lung volume (3.21 ± 0.42% vs. 3.48 ± 0.38%; P = 5.0 × 10-12), lower ratios of airway to lung volume (0.031 ± 0.003 vs. 0.034 ± 0.004; P = 6.1 × 10-13), and larger ratios of small vessel volume to TPVV (37.91 ± 4.26% vs. 35.53 ± 4.89%; P = 1.9 × 10-7). In adjusted analyses, an increment of 1 standard deviation in PAARseg was associated with a 98.4% higher rate of severe exacerbations (95% confidence interval, 29-206%; P = 0.002) and 79.3% higher odds of being in the RP group (95% confidence interval, 24-157%; P = 0.001). At 2-year follow-up, the CT-defined RP group demonstrated a significant decline in postbronchodilator percentage predicted forced expiratory volume in 1 second. Conclusions: Rapid one-year progression of emphysema was associated with indices indicative of higher peripheral pulmonary vascular resistance and a possible role played by pulmonary vascular-airway dysanapsis.

A Survey on Artificial Intelligence in Pulmonary Imaging.

Over the last decade, deep learning (DL) has contributed a paradigm shift in computer vision and image recognition creating widespread opportunities of using artificial intelligence in research as well as industrial applications. DL has been extensively studied in medical imaging applications, including those related to pulmonary diseases. Chronic obstructive pulmonary disease, asthma, lung cancer, pneumonia, and, more recently, COVID-19 are common lung diseases affecting nearly 7.4% of world population. Pulmonary imaging has been widely investigated toward improving our understanding of disease etiologies and early diagnosis and assessment of disease progression and clinical outcomes. DL has been broadly applied to solve various pulmonary image processing challenges including classification, recognition, registration, and segmentation. This paper presents a survey of pulmonary diseases, roles of imaging in translational and clinical pulmonary research, and applications of different DL architectures and methods in pulmonary imaging with emphasis on DL-based segmentation of major pulmonary anatomies such as lung volumes, lung lobes, pulmonary vessels, and airways as well as thoracic musculoskeletal anatomies related to pulmonary diseases.

Continuum finite element analysis generalizesin vivotrabecular bone microstructural strength measures between two CT scanners with different image resolution.

Fragility of trabecular bone (Tb) microstructure is increased in osteoporosis, which is associated with rapid bone loss and enhanced fracture-risk. Accurate assessment of Tb strength usingin vivoimaging available in clinical settings will be significant for management of osteoporosis and understanding its pathogenesis. Emerging CT technology, featured with high image resolution, fast scan-speed, and wide clinical access, is a promising alternative forin vivoTb imaging. However, variation in image resolution among different CT scanners pose a major hurdle in CT-based bone studies. This paper presents nonlinear continuum finite element (FE) methods for computation of Tb strength fromin vivoCT imaging and evaluates their generalizability between two scanners with different image resolution. Continuum FE-based measures of Tb strength under different loading conditions were found to be highly reproducible (ICC ≥ 0.93) using ankle images of twenty healthy volunteers acquired on low- and high-resolution CT scanners 44.6 ± 2.7 days apart. FE stress propagation was mostly confined to Tb micro-network (2.3 ± 1.7 MPa) with nominal leakages over the marrow space (0.4 ± 0.5 MPa) complying with the fundamental principle of mechanics atin vivoimaging. In summary, nonlinear continuum FE-based Tb strength measures are reproducible among different CT scanners and suitable for multi-site longitudinal human studies.

Associations Between Muscle Weakness and Clinical Outcomes in Current and Former Smokers.

Introduction: Smokers with chronic obstructive pulmonary disease (COPD) are at increased risk of muscle weakness. There are limited data describing weakness in smokers with normal spirometry and preserved ratio-impaired spirometry (PRISm), 2 subgroups at risk of respiratory symptom burden and activity limitations. In this study, we evaluated the associations of 2 weakness measures, sit-to-stand (STS) and handgrip strength (HGS), with clinical outcomes in smokers with COPD, normal spirometry, and PRISm. Methods: We evaluated 1972 current and former smokers from the COPD Genetic Epidemiology (COPDGene®) cohort with STS and HGS measurements at their 10-year study visit. Multivariable regression modeling was used to assess associations between weakness measures and the 6-minute walk distance (6MWD) test, the St George's Respiratory Questionnaire (SGRQ), the Short-Form-36 (SF-36), severe exacerbations, and prospective mortality, reported as standardized coefficients (ß), odds ratios (ORs), or hazard ratios (HRs). Results: Compared with HGS, STS was more strongly associated with the 6MWD (ß=0.45, p<0.001 versus. ß=0.25, p<0.001), SGRQ (ß=-0.24, p<0.001 versus ß=-0.18, p<0.001), SF-36 Physical Functioning (ß=0.36, p<0.001 versus ß=0.25, p<0.001), severe exacerbations (OR 0.95, p=0.04 versus OR 0.97, p=0.01), and prospective mortality (HR 0.83, p=0.001 versus HR 0.94, p=0.03). Correlations remained after stratification by spirometric subgroups. Compared with males, females had larger magnitude effect sizes between STS and clinical outcomes. Conclusions: STS and HGS are easy to perform weakness measures that provide important information about functional performance, health-related quality of life, severe exacerbations, and survival in smokers, regardless of spirometric subgroup. This iterates the importance of screening current and former smokers for weakness in the outpatient setting.

A new multi-object image thresholding method based on correlation between object class uncertainty and intensity gradient.

PURPOSE: Image thresholding and gradient analysis have remained popular image preprocessing tools for several decades due to the simplicity and straight-forwardness of their definitions. Also, optimum selection of threshold and gradient strength values are hidden steps in many advanced medical imaging algorithms. A reliable method for threshold optimization may be a crucial step toward automation of several medical image based applications. Most automatic thresholding and gradient selection methods reported in literature primarily focus on image histograms ignoring a significant amount of information embedded in the spatial distribution of intensity values forming visible features in an image. Here, we present a new method that simultaneously optimizes both threshold and gradient values for different object interfaces in an image that is based on unification of information from both the histogram and spatial image features; also, the method works for unknown number of object regions. METHODS: A new energy function is formulated by combining the object class uncertainty measure, a histogram-based feature, of each pixel with its image gradient measure, a spatial contextual feature in an image. The energy function is designed to measure the overall compliance of the theoretical premise that, in a probabilistic sense, image intensities with high class uncertainty are associated with high image gradients. Finally, it is expressed as a function of threshold and gradient parameters and optimum combinations of these parameters are sought by locating pits and valleys on the energy surface. A major strength of the algorithm lies in the fact that it does not require the number of object regions in an image to be predefined. RESULTS: The method has been applied on several medical image datasets and it has successfully determined both threshold and gradient parameters for different object interfaces even when some of the thresholds are almost impossible to locate in the histogram. Both accuracy and reproducibility of the method have been examined on several medical image datasets including repeat scan 3D multidetector computed tomography (CT) images of cadaveric ankles specimens. Also, the new method has been qualitatively and quantitatively compared with Otsu's method along with three other algorithms based on minimum error thresholding, maximum segmented image information and minimization of homogeneity- and uncertainty-based energy and the results have demonstrated superiority of the new method. CONCLUSIONS: We have developed a new automatic threshold and gradient strength selection algorithm by combining class uncertainty and spatial image gradient features. The performance of the method has been examined in terms of accuracy and reproducibility and the results found are better as compared to several popular automatic threshold selection methods.