Conquering the Cobb Angle: A Deep Learning Algorithm for Automated, Hardware-Invariant Measurement of Cobb Angle on Radiographs in Patients with Scoliosis.

Scoliosis is a disease estimated to affect more than 8% of adults in the United States. It is diagnosed with use of radiography by means of manual measurement of the angle between maximally tilted vertebrae on a radiograph (ie, the Cobb angle). However, these measurements are time-consuming, limiting their use in scoliosis surgical planning and postoperative monitoring. In this retrospective study, a pipeline (using the SpineTK architecture) was developed that was trained, validated, and tested on 1310 anterior-posterior images obtained with a low-dose stereoradiographic scanning system and radiographs obtained in patients with suspected scoliosis to automatically measure Cobb angles. The images were obtained at six centers (2005-2020). The algorithm measured Cobb angles on hold-out internal (n = 460) and external (n = 161) test sets with less than 2° error (intraclass correlation coefficient, 0.96) compared with ground truth measurements by two experienced radiologists. Measurements, produced in less than 0.5 second, did not differ significantly (P = .05 cutoff) from ground truth measurements, regardless of the presence or absence of surgical hardware (P = .80), age (P = .58), sex (P = .83), body mass index (P = .63), scoliosis severity (P = .44), or image type (low-dose stereoradiographic image vs radiograph; P = .51) in the patient. These findings suggest that the algorithm is highly robust across different clinical characteristics. Given its automated, rapid, and accurate measurements, this network may be used for monitoring scoliosis progression in patients. Keywords: Cobb Angle, Convolutional Neural Network, Deep Learning Algorithms, Pediatrics, Machine Learning Algorithms, Scoliosis, Spine Supplemental material is available for this article. © RSNA, 2023.

Unsolicited Invitations to Scientific Meetings: Radiologists' Experience.

OBJECTIVE: Unsolicited invitations to speak at medical meetings have proliferated as a type of spam email and phishing strategy to scam unsuspecting victims. We sought to determine the prevalence of such invitations to questionable meetings and determine the factors associated with receiving such solicitations. MATERIALS AND METHODS: Data was collected for the number of speakers' invitations received over a 2-week period (April -May 2023) by radiologists of different subspecialties, academic ranks, and histories of publications and speaking engagements in the past 2-5 years. We analyzed the number of invitations received based on the variables. RESULTS: Thirty-three of 45 (73.3%) faculty members received 188 inappropriate invitation emails in the 2-week observation period. The mean number of invitation emails was 4.13 for each faculty (SD: 5.03, range 0-20). There was no correlation between the number of invitations and radiologists' subspecialty, academic rank (3.8 ± 5, 3.0 ± 4, and 5.5 ± 5.7 invitations for full, associate, and assistant professors respectively) and previous legitimate speaker invites. Only 6 (3.2%) out of 188 invitations to speak sent to radiologists were for radiology-related meetings. Having more than 10 publications since 2022 was associated with a 5.0 (1.2, 19.4) times higher odds of receiving more than 4 solicitations. CONCLUSIONS: A total of 73.3% of the faculty surveyed received unsolicited invitations to meetings in the 2-week study period and over 96% of the invitations were unrelated to their field of practice. Our results show that publications since 2022 was the most significant factor associated with receiving more solicitations. CLINICAL RELEVANCE/APPLICATION: Invitations to questionable meetings targeting radiologists are frequent and often are unrelated to their specialties. The risk factors for receiving the invitations are unclear. Understanding these risk factors may enable educators especially junior investigators, to be better prepared to appropriately address such solicitations.

Vertebral Deformity Measurements at MRI, CT, and Radiography Using Deep Learning.

PURPOSE: To construct and evaluate the efficacy of a deep learning system to rapidly and automatically locate six vertebral landmarks, which are used to measure vertebral body heights, and to output spine angle measurements (lumbar lordosis angles [LLAs]) across multiple modalities. MATERIALS AND METHODS: In this retrospective study, MR (n = 1123), CT (n = 137), and radiographic (n = 484) images were used from a wide variety of patient populations, ages, disease stages, bone densities, and interventions (n = 1744 total patients, 64 years ± 8, 76.8% women; images acquired 2005-2020). Trained annotators assessed images and generated data necessary for deformity analysis and for model development. A neural network model was then trained to output vertebral body landmarks for vertebral height measurement. The network was trained and validated on 898 MR, 110 CT, and 387 radiographic images and was then evaluated or tested on the remaining images for measuring deformities and LLAs. The Pearson correlation coefficient was used in reporting LLA measurements. RESULTS: On the holdout testing dataset (225 MR, 27 CT, and 97 radiographic images), the network was able to measure vertebral heights (mean height percentage of error ± 1 standard deviation: MR images, 1.5% ± 0.3; CT scans, 1.9% ± 0.2; radiographs, 1.7% ± 0.4) and produce other measures such as the LLA (mean absolute error: MR images, 2.90°; CT scans, 2.26°; radiographs, 3.60°) in less than 1.7 seconds across MR, CT, and radiographic imaging studies. CONCLUSION: The developed network was able to rapidly measure morphometric quantities in vertebral bodies and output LLAs across multiple modalities.Keywords: Computer Aided Diagnosis (CAD), MRI, CT, Spine, Demineralization-Bone, Feature Detection Supplemental material is available for this article. © RSNA, 2021.

Evaluation of the impact of age and adiposity on a multi-biomarker disease activity score before and after adjustment.

PURPOSE: We assessed the impact of adjustment of the multi-biomarker disease activity score (MBDA) for age, sex, and leptin, over the range of age and adiposity, and assessed relationships with clinical disease activity. METHODS: Patients with RA, ages 18-75 years, were recruited from clinical practices and completed whole-body DXA to quantify fat mass indices (FMI, kg/m2). FMI Z-scores were calculated based on distributions in a reference population. Descriptive statistics described relationships between age, FMI Z-score, and the original MBDA and adjusted MBDA (aMBDA). Swollen joint counts (SJC) and the clinical disease activity index (CDAI) were assessed over MBDA categories. RESULTS: There were 104 participants (50% female) with mean (SD) age of 56.1 (12.5) and body mass index (BMI) of 28.8 (6.9). Older age was associated with higher MBDA scores in men. The aMBDA was not associated with age. The original MBDA score was associated with FMI Z-score among women (Rho = 0.42, p = 0.002) but not men. The aMBDA was not associated with FMI Z-score in either women or men. The aMBDA score was lower than the original MBDA in the highest quartile of FMI in women and was higher in the lowest FMI quartiles in women and men. CDAI, SJC, and radiographic scores were similar across activity categories for the original MBDA score and aMBDA. CONCLUSIONS: The aMBDA demonstrated reduced associations with adiposity, particularly among women. The aMBDA may be less likely to overestimate disease activity in women with greater adiposity and to underestimate disease activity in men and women with lesser adiposity. Key Points • Leptin adjustment of the MBDA score reduces the influence of adiposity, particularly among women. • Leptin adjustment results in significantly higher estimated disease activity in thin men and women. • The adjusted and unadjusted score correlate similarly with clinical disease activity measures.

Association of Low Muscle Density With Deteriorations in Muscle Strength and Physical Functioning in Rheumatoid Arthritis.

OBJECTIVE: Rheumatoid arthritis (RA) is associated with low muscle density due to the accumulation of intramuscular fat. The present study was undertaken to identify predictors of changes in muscle density and to determine whether low muscle density predicted changes in strength and physical function. METHODS: Patients with RA, ages 18-70 years, completed whole-body dual-energy x-ray absorptiometry and peripheral quantitative computed tomography to quantify lean and fat mass indices and muscle density. Dynamometry was used to measure strength at the hand, knee, and lower leg. Disability and physical function were measured with the Health Assessment Questionnaire (HAQ) and the Short Physical Performance Battery (SPPB). Assessments were performed at baseline and at follow-up. Regression analyses assessed associations between patient characteristics, muscle density, and deteriorations in strength and function. RESULTS: Muscle density was assessed at baseline in 107 patients with RA. Seventy-nine of these patients (74%) returned for a follow-up assessment at a median follow-up time of 2.71 years (interquartile range 2.35-3.57). Factors associated with declines in muscle density included female sex, higher disease activity, smoking, and lower insulin-like growth factor 1 (IGF-1) levels. Greater muscle density Z score at baseline (per 1 SD) was associated with less worsening per year according to HAQ, SPPB, and 4-meter walk time scores and a lower risk of a clinically important worsening in HAQ score (odds ratio [OR] 1.90 [95% confidence interval (95% CI) 1.06, 3.42]; P = 0.03) and walking speed (OR 2.87 [95% CI 1.05, 7.89]; P = 0.04). CONCLUSION: Worsening of skeletal muscle density occurred in patients with higher disease activity, in smokers, and in those with lower IGF-1. Low muscle density was associated with worsening of physical function. Interventions addressing reductions in muscle quality might prevent functional decline.

Circulating Fibroblast Growth Factor-21 Levels in Rheumatoid Arthritis: Associations With Disease Characteristics, Body Composition, and Physical Functioning.

OBJECTIVE: This study evaluated associations between fibroblast growth factor (FGF)-21, an adipokine associated with metabolic stress, and adverse longitudinal changes in body composition and physical functioning in patients with rheumatoid arthritis (RA). METHODS: At baseline and follow-up, patients with RA aged 18-70 years completed whole-body dual-energy X-ray absorptiometry and peripheral quantitative computed tomography to quantify lean mass, fat mass, and muscle density. Dynamometry assessed muscle strength at the hand and knee, and physical functioning was measured with the Health Assessment Questionnaire (HAQ) and the Short Physical Performance Battery (SPPB). FGF-21 and inflammatory cytokines were measured at baseline. Linear and logistic regression analyses assessed associations between FGF-21 levels and both body composition and physical functioning over time. RESULTS: There were 113 patients with RA enrolled, and 84 (74%) returned for follow-up at a median of 2.68 years. At baseline, FGF-21 was associated with age, smoking, methotrexate use, adiposity, and inflammatory cytokines: tumor necrosis factor receptor type I, YKL-40, vascular endothelial growth factor (VEGF), and resistin. The highest FGF-21 quartile was associated with worse SPPB and HAQ. Higher baseline FGF-21 levels (per 1 SD) were associated with worsening in muscle density and area Z-scores (ß -0.06, 95% CI -0.12 to 0.008, P = 0.08; and ß -0.05, 95% CI -0.10 to 0.006, P = 0.08, respectively) and a greater probability of a clinically meaningful worsening of HAQ (OR 2.37, 95% CI 1.21-4.64, P = 0.01). The fourth FGF-21 quartile was associated with worsening of SPPB (ß -0.57, 95% CI -1.04 to -0.09, P = 0.02). CONCLUSION: FGF-21 levels are associated with obesity and inflammatory cytokines, and with worsening in physical functioning in RA. These data support the hypothesis that FGF-21 can identify patients at risk of functional decline.

A novel method to assess subchondral bone formation using [18F]NaF-PET in the evaluation of knee degeneration.

PURPOSE: Fluorine-18-sodium fluoride-PET ([F]NaF-PET) facilitates direct assessment of subchondral bone formation to evaluate degeneration in articulating joints. No standards exist for the quantification of joint activity using [F]NaF-PET, and many techniques rely on focal uptake to characterize an entire region of interest. This study proposes a novel method of quantitative global knee analysis to assess regions of expected bone remodeling in the evaluation of knee degeneration. PATIENTS AND METHODS: The study population consisted of 18 patients with rheumatoid arthritis who underwent [F]NaF-PET/computed tomography imaging. The maximum standardized uptake value (knee SUVmax) in addition to a target-to-background ratio (TBR) that represents global knee activity adjusted for systemic bone formation measured at the lateral femoral neck (global knee TBR) were calculated. A radiologist scored standard radiographs of the knee in nine patients using the Kellgren-Lawrence grading system. RESULTS: Patients with greater [F]NaF uptake demonstrated greater knee deterioration, which was corroborated by the radiograph findings. Average Kellgren-Lawrence grading was strongly associated with both global knee TBR (Spearman ρ=0.69, P=0.04) and knee SUVmax scores (Spearman ρ=0.93, P=0.0003). CONCLUSION: Assessment of global activity within the joint is a feasible and clinically useful technique for characterizing disease activity with a single value. Furthermore, a ratio based on systemic bone turnover in a nonarticulating, weight-bearing site adjusts for differences in bone formation related to bodyweight or metabolic bone diseases. We hypothesize that a global knee TBR score may be more sensitive at detecting changes in disease progression, as new spatially distinct lesions with a lower SUV that develop within an region of interest would not be detected by the SUVmax methodology. Longitudinal studies assessing sensitivity with larger patient cohorts are needed to further validate this methodology.

Intramuscular Fat Accumulation and Associations With Body Composition, Strength, and Physical Functioning in Patients With Rheumatoid Arthritis.

OBJECTIVE: Rheumatoid arthritis (RA) is associated with adverse body composition profiles and low muscle density due to the accumulation of intramuscular fat. Linear regression was used to assess differences between RA patients and controls and to determine associations between muscle density, strength, and physical functioning. METHODS: Patients with RA, ages 18-70 years, and healthy control subjects underwent whole-body dual x-ray absorptiometry and peripheral quantitative computed tomography, in order to quantify the appendicular lean mass index (ALMI) and the fat mass index (FMI), visceral fat area, and muscle density. Dynamometry was used to measure hand grip strength and muscle strength at the knee and lower leg. Disability and physical functioning were measured using the Health Assessment Questionnaire (HAQ) and the Short Physical Performance Battery (SPPB). Linear regression analyses were performed to assess differences related to RA and associations between muscle density, strength, and function. RESULTS: The study group included 103 patients with RA (51 men) and 428 healthy control subjects. Among patients with RA, low muscle density was associated with higher disease activity, C-reactive protein and interleukin-6 levels, greater total and visceral fat area, lower ALMI Z scores, physical inactivity, and long-term use of glucocorticoids (>1 year). Patients with low ALMI Z scores had lower muscle density Z scores compared with reference participants with similarly low ALMI scores. Low muscle density was independently associated with lower muscle strength, higher HAQ scores, and lower SPPB scores, after adjustment for ALMI and FMI Z scores. CONCLUSION: The low muscle density observed in patients with RA was associated with low muscle mass, excess adiposity, poor strength, and greater disability. Interventions to address poor muscle quality could potentially affect important functional outcomes.

Muscle Deficits in Rheumatoid Arthritis Contribute to Inferior Cortical Bone Structure and Trabecular Bone Mineral Density.

OBJECTIVE: Rheumatoid arthritis (RA) is associated with muscle loss, osteoporosis, and fracture. We examined associations between skeletal muscle mass, strength, and quality and trabecular and cortical bone deficits in patients with RA and healthy controls. METHODS: Participants, ages 18-75 years, completed whole-body dual-energy x-ray absorptiometry and peripheral quantitative computed tomography (pQCT) of the tibia to quantify appendicular lean mass and fat mass indices (ALMI, FMI), muscle density at the lower leg, trabecular bone density, and cortical bone thickness. Age-, sex-, and race-specific Z scores were calculated based on distributions in controls. Associations between body composition and pQCT bone outcomes were assessed in patients with RA and controls. Linear regression analyses assessed differences in bone outcomes after considering differences in body mass index (BMI) and body composition. RESULTS: The sample consisted of 112 patients with RA (55 men) and 412 controls (194 men). Compared to controls, patients with RA had greater BMI Z score (p < 0.001), lower ALMI Z score after adjustment for FMI (p = 0.02), lower muscle strength Z score (p = 0.01), and lower muscle density Z score (p < 0.001). Among RA, ALMI Z scores were positively associated with trabecular density [ß: 0.29 (0.062-0.52); p = 0.01] and cortical thickness [ß: 0.33 (0.13-0.53; p = 0.002]. Associations were similar in controls. Bone outcomes were inferior in patients with RA after adjusting for BMI, but similar to controls when adjusting for body composition. Radiographic damage and higher adiponectin levels were independently associated with inferior bone outcomes. CONCLUSION: Patients with RA exhibit deficits in cortical bone structure and trabecular density at the tibia and a preserved functional muscle-bone unit. A loss of mechanical loading may contribute to bone deficits.

Insulin-like Growth Factor 1 and Adiponectin and Associations with Muscle Deficits, Disease Characteristics, and Treatments in Rheumatoid Arthritis.

OBJECTIVE: Rheumatoid arthritis (RA) is associated with low muscle mass and density. The objective of our study was to evaluate associations between 2 serum biomarkers [insulin-like growth factor 1 (IGF-1) and adiponectin] and skeletal muscle in RA. METHODS: Whole-body dual energy X-ray absorptiometry measures of the appendicular lean mass index (ALMI; kg/m(2)) and total fat mass index (kg/m(2)), as well as the peripheral quantitative computed tomography measures of the lower leg muscle and fat cross-sectional area (CSA; cm(2)) and muscle density (an index of fat infiltration) were obtained from 50 participants with RA, ages 18-70 years. Multivariable linear regression analyses evaluated associations between body composition and levels of adiponectin and IGF-1, adjusted for age, sex, and adiposity. RESULTS: Greater age was associated with higher adiponectin (p = 0.06) and lower IGF-1 (p = 0.004). Eight subjects had IGF-1 levels below the reference range for their age and sex. These subjects had significantly lower ALMI and muscle CSA in multivariable models. Lower IGF-1 levels were associated with greater clinical disease activity and severity, as well as low ALMI, muscle CSA, and muscle density (defined as 1 SD below normative mean). After adjusting for age and sex, greater adiponectin levels were associated with lower BMI (p = 0.02) as well as lower ALMI, and lower muscle CSA, independent of adiposity (p < 0.05). Only greater Health Assessment Questionnaire scores were significantly associated with lower adiponectin levels. CONCLUSION: Low IGF-1 and greater adiponectin levels are associated with lower muscle mass in RA. Lower IGF-1 levels were seen in subjects with greater disease activity and severity.

Deficits in muscle mass, muscle density, and modified associations with fat in rheumatoid arthritis.

OBJECTIVE: To quantify muscle outcomes, independent of fat mass, in rheumatoid arthritis (RA) patients compared to healthy controls. METHODS: Quantitative computed tomography scans measured calf muscle and fat cross-sectional area (CSA) and muscle density (an index of intramuscular adipose tissue), and isometric dynamometry was used to measure ankle muscle strength in 50 participants with RA ages 18-70 years and 500 healthy controls. Multivariable linear regression models assessed muscle deficits in RA after adjusting for group differences in adiposity and assessing for an altered muscle-fat association. Associations between RA disease characteristics and fat-adjusted muscle outcomes were also assessed. RESULTS: Compared to controls, RA subjects had significantly greater body mass index (BMI) and fat area, and lower muscle area, muscle density, and muscle strength (P < 0.001 for all). Strength deficits were eliminated with adjustment for the smaller muscle area. The magnitude of muscle deficits, relative to controls, was significantly greater (P < 0.03 for interaction) in participants with lower fat area and BMI. Among those in the lower tertiles of adiposity, RA subjects demonstrated more significant deficits compared to controls with similar adiposity. In contrast, among those in the highest tertile for adiposity, RA was not associated with muscle deficits. Among RA, greater Sharp/van der Heijde scores were associated with lower muscle CSA and muscle density. Greater disease activity and disability were associated with low muscle density. CONCLUSION: Deficits in muscle area and muscle density are present in RA patients compared to controls and are most pronounced in subjects with low fat mass. Greater joint destruction is associated with greater muscle deficits.