RESUMO
The Convolutional Neural Network algorithm achieved a sensitivity of 94% and specificity of 93% in identifying scans with vertebral fractures (VFs). The external validation results suggest that the algorithm provides an opportunity to aid radiologists with the early identification of VFs in routine CT scans of abdomen and chest. PURPOSE: To evaluate the performance of a previously trained Convolutional Neural Network (CNN) model to automatically detect vertebral fractures (VFs) in CT scans in an external validation cohort. METHODS: Two Chinese studies and clinical data were used to retrospectively select CT scans of the chest, abdomen and thoracolumbar spine in men and women aged ≥50 years. The CT scans were assessed using the semiquantitative (SQ) Genant classification for prevalent VFs in a process blinded to clinical information. The performance of the CNN model was evaluated against reference standard readings by the area under the receiver operating characteristics curve (AUROC), accuracy, Cohen's kappa, sensitivity, and specificity. RESULTS: A total of 4,810 subjects were included, with a median age of 62 years (IQR 56-67), of which 2,654 (55.2%) were females. The scans were acquired between January 2013 and January 2019 on 16 different CT scanners from three different manufacturers. 2,773 (57.7%) were abdominal CTs. A total of 628 scans (13.1%) had ≥1 VF (grade 2-3), representing 899 fractured vertebrae out of a total of 48,584 (1.9%) visualized vertebral bodies. The CNN's performance in identifying scans with ≥1 moderate or severe fractures achieved an AUROC of 0.94 (95% CI: 0.93-0.95), accuracy of 93% (95% CI: 93%-94%), kappa of 0.75 (95% CI: 0.72-0.77), a sensitivity of 94% (95% CI: 92-96%) and a specificity of 93% (95% CI: 93-94%). CONCLUSION: The algorithm demonstrated excellent performance in the identification of vertebral fractures in a cohort of chest and abdominal CT scans of Chinese patients ≥50 years.
Assuntos
Fraturas da Coluna Vertebral , Masculino , Humanos , Feminino , Pessoa de Meia-Idade , Idoso , Estudos Retrospectivos , Fraturas da Coluna Vertebral/diagnóstico por imagem , Tomografia Computadorizada por Raios X/métodos , Algoritmos , Redes Neurais de ComputaçãoRESUMO
In men and women with opportunistically identifiable vertebral fractures (VFs) on routine CT scans including the chest and/or abdomen, the risk of death is 51% higher than in those with no VF on the CT scan, and 325% higher than an age- and sex-matched general population cohort. PURPOSE: There is little knowledge about the risk of death in patients with VFs present on routine radiological imaging. We evaluated the risk of death in men and women aged 50 years or older with opportunistically identifiable VFs on routine CT scans and not treated with osteoporosis medications. METHODS: Thoracic and lumbar VFs were identified through a blinded, two-step approach on CT scans performed as part of normal clinical care in a Danish hospital in 2010 or later. Subjects with VF were matched on age and sex against those with no VF (1:2-ratio) and a general population cohort (1:3-ratio), respectively, and followed for up to 7 years through the national Danish registers. Subjects treated with an osteoporosis medication in the year prior to baseline were excluded. RESULTS: Subjects with VF had a significantly higher risk of death during follow-up as compared to subjects with no VF on the CT scan (adjusted hazard ratio [HR] 1.51 [95% confidence interval 1.27-1.79; p < 0.001]) and even more so when compared to the general population cohort (HR 4.25 [3.53-5.12; p < 0.001]). In subjects with versus without VF on the CT scan, the risk was higher in those with moderate or severe VF, in those with no malignancy prior to baseline, and in those with a lower Charlson comorbidity index score. CONCLUSION: Subjects with VF available for identification on routine CT scans face a substantially increased risk of death. Opportunistic identification and reporting of VF is important to identify these patients to allow intervention if indicated.
Assuntos
Osteoporose , Fraturas por Osteoporose , Fraturas da Coluna Vertebral , Feminino , Humanos , Masculino , Densidade Óssea , Estudos de Coortes , Osteoporose/epidemiologia , Fraturas por Osteoporose/epidemiologia , Fraturas da Coluna Vertebral/epidemiologia , Tomografia Computadorizada por Raios X/métodos , Pessoa de Meia-IdadeRESUMO
PURPOSE: Vertebral fractures (VFs) are often available on radiological imaging undertaken during daily clinical work, yet the healthcare cost burden of these opportunistically identifiable fractures has not previously been reported. In this study, we examine the direct healthcare costs of subjects with vertebral fractures available for identification on routine CT scans. METHODS: Thoracolumbar vertebral fractures were identified from 2000 routine CT scans. Subjects with VF on the scan were matched 1:2 against subjects with no VF on the scan, and similarly in a 1:3-ratio against a general population cohort. We excluded those subjects who received treatment with osteoporosis medication(s) in the year prior to baseline. Direct healthcare costs, identified from the national Danish registers, were accrued over up to 6 years of follow-up, and reported per day at risk and per year. RESULTS: In subjects undergoing a CT scan, costs were initially high, yet declined over time. Comparing subjects with prevalent vertebral fracture (n = 321) against those subjects with no vertebral fracture (n = 606), mean total healthcare costs per day at risk was numerically higher in the first three years after baseline, while healthcare costs per year were similar between the cohorts. No differences reached statistical significance. When compared to the general population cohort, costs were significantly higher in the vertebral fracture cohort. CONCLUSION: Subjects with vertebral fractures available for identification on routine CT scans incur substantially higher healthcare costs than matched subjects representing the general population, and numerically, albeit non-significantly, higher healthcare costs per day at risk in the short term, as compared to subjects with no visible VF on the CT scan.
Assuntos
Osteoporose , Fraturas por Osteoporose , Fraturas da Coluna Vertebral , Humanos , Fraturas por Osteoporose/diagnóstico por imagem , Fraturas por Osteoporose/epidemiologia , Osteoporose/complicações , Fraturas da Coluna Vertebral/diagnóstico por imagem , Fraturas da Coluna Vertebral/epidemiologia , Fraturas da Coluna Vertebral/complicações , Custos de Cuidados de Saúde , Densidade ÓsseaRESUMO
Vertebral fractures (VFs) are the hallmark of osteoporosis, being one of the most frequent types of fragility fracture and an early sign of the disease. They are associated with significant morbidity and mortality. VFs are incidentally found in one out of five imaging studies, however, more than half of the VFs are not identified nor reported in patient computed tomography (CT) scans. Our study aimed to develop a machine learning algorithm to identify VFs in abdominal/chest CT scans and evaluate its performance. We acquired two independent data sets of routine abdominal/chest CT scans of patients aged 50 years or older: a training set of 1011 scans from a non-interventional, prospective proof-of-concept study at the Universitair Ziekenhuis (UZ) Brussel and a validation set of 2000 subjects from an observational cohort study at the Hospital of Holbaek. Both data sets were externally reevaluated to identify reference standard VF readings using the Genant semiquantitative (SQ) grading. Four independent models have been trained in a cross-validation experiment using the training set and an ensemble of four models has been applied to the external validation set. The validation set contained 15.3% scans with one or more VF (SQ2-3), whereas 663 of 24,930 evaluable vertebrae (2.7%) were fractured (SQ2-3) as per reference standard readings. Comparison of the ensemble model with the reference standard readings in identifying subjects with one or more moderate or severe VF resulted in an area under the receiver operating characteristic curve (AUROC) of 0.88 (95% confidence interval [CI], 0.85-0.90), accuracy of 0.92 (95% CI, 0.91-0.93), kappa of 0.72 (95% CI, 0.67-0.76), sensitivity of 0.81 (95% CI, 0.76-0.85), and specificity of 0.95 (95% CI, 0.93-0.96). We demonstrated that a machine learning algorithm trained for VF detection achieved strong performance on an external validation set. It has the potential to support healthcare professionals with the early identification of VFs and prevention of future fragility fractures. © 2023 UCB S.A. and The Authors. Journal of Bone and Mineral Research published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research (ASBMR).
Assuntos
Fraturas por Osteoporose , Fraturas da Coluna Vertebral , Humanos , Estudos Prospectivos , Fraturas da Coluna Vertebral/diagnóstico por imagem , Fraturas da Coluna Vertebral/complicações , Tomografia Computadorizada por Raios X/métodos , Algoritmos , Aprendizado de Máquina , Minerais , Fraturas por Osteoporose/diagnóstico por imagem , Fraturas por Osteoporose/complicações , Densidade ÓsseaRESUMO
Vertebral fractures (VFs) have been associated with future fractures, yet few studies have evaluated whether this pertains to VFs available for identification on routine radiological imaging. We sought to evaluate the risk of subsequent fractures in subjects with VF identified opportunistically on computed tomography (CT) scans performed as part of routine clinical practice. From the radiology database of Holbæk Hospital we identified the first CT scan including the thorax and/or abdomen of 2000 consecutive men and women aged 50 years or older, performed from January 1, 2010 onward. The scans were assessed in a blinded approach to identify chest and lumbar VF, and these data linked to national Danish registers. Subjects were excluded if treated with an osteoporosis medication (OM) in the year prior to baseline (date of CT), and the remaining subjects with VF matched on age and sex in 1:2 ratio against subjects with no VF. We found that the risk of major osteoporotic fractures (hip, non-cervical vertebral, humerus, and distal forearm fractures) was higher for subjects with VF than without VF: incidence rates (IRs) were 32.88 and 19.59 fractures per 1000 subject-years, respectively, and the adjusted hazard ratio (HRadj) was 1.72 (95% confidence interval [CI], 1.03-2.86). Subsequent hip fracture IRs were 16.75 and 6.60; HRadj 3.02 (95% CI, 1.39-6.55). There were no significant differences in other fracture outcomes (including a pooled estimate of any subsequent fracture, except face, skull, and fingers: IRs 41.52 and 31.38; HRadj 1.31 [95% CI, 0.85-2.03]). Our findings suggest that subjects undergoing routine CT scans including the chest and/or abdomen are a high risk population in terms of fracture risk. Even within this group, subjects with VF are at higher risk of future major osteoporotic fracture (MOF), in particular hip fracture. Hence, systematic opportunistic screening for VF and subsequent fracture risk management is important to reduce the risk of new fractures. © 2023 The Authors. JBMR Plus published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research.