Patients with anorexia nervosa had increased skewness and kurtosis and decreased entropy and mean value of positive pixels compared with normal-weight control subjects, and these parameters were associated with lowest lifetime weight and duration of amenorrhea; there were no such associations with bone mineral density, and these findings suggest that trabecular texture analysis might contribute information about bone health in anorexia nervosa that is independent of information provided with bone mineral density.
Abstract
Purpose
To determine indexes of skeletal integrity by using computed tomographic (CT) trabecular texture analysis of the lumbar spine in patients with anorexia nervosa and normal-weight control subjects and to determine body composition predictors of trabecular texture.
Materials and Methods
This cross-sectional study was approved by the institutional review board and compliant with HIPAA. Written informed consent was obtained. The study included 30 women with anorexia nervosa (mean age ± standard deviation, 26 years ± 6) and 30 normal-weight age-matched women (control group). All participants underwent low-dose single-section quantitative CT of the L4 vertebral body with use of a calibration phantom. Trabecular texture analysis was performed by using software. Skewness (asymmetry of gray-level pixel distribution), kurtosis (pointiness of pixel distribution), entropy (inhomogeneity of pixel distribution), and mean value of positive pixels (MPP) were assessed. Bone mineral density and abdominal fat and paraspinal muscle areas were quantified with quantitative CT. Women with anorexia nervosa and normal-weight control subjects were compared by using the Student t test. Linear regression analyses were performed to determine associations between trabecular texture and body composition.
Results
Women with anorexia nervosa had higher skewness and kurtosis, lower MPP (P < .001), and a trend toward lower entropy (P = .07) compared with control subjects. Bone mineral density, abdominal fat area, and paraspinal muscle area were inversely associated with skewness and kurtosis and positively associated with MPP and entropy. Texture parameters, but not bone mineral density, were associated with lowest lifetime weight and duration of amenorrhea in anorexia nervosa.
Conclusion
Patients with anorexia nervosa had increased skewness and kurtosis and decreased entropy and MPP compared with normal-weight control subjects. These parameters were associated with lowest lifetime weight and duration of amenorrhea, but there were no such associations with bone mineral density. These findings suggest that trabecular texture analysis might contribute information about bone health in anorexia nervosa that is independent of that provided with bone mineral density.
© RSNA, 2016
Introduction
Anorexia nervosa is a prevalent eating disorder associated with low bone mineral density (BMD) and impaired bone microarchitecture and bone strength, resulting in increased fracture risk (1–4). Skeletal sites containing predominantly trabecular bone, such as the lumbar spine, are the most severely affected, and patients with anorexia nervosa have an increased risk for spinal compression fractures (5,6).
Dual-energy x-ray absorptiometry (DXA) is most commonly used to assess BMD; however, it is influenced by body size (7), which provides a diagnostic challenge in anorexia nervosa because prolonged undernutrition can lead to reduced muscle and fat mass. In addition, bone microarchitecture affects bone strength independent of BMD. Studies examining the distal radius in female patients with anorexia nervosa have shown impaired trabecular and cortical microarchitecture (as assessed with high-spatial-resolution flat-panel computed tomography [CT] [1] and high-spatial-resolution peripheral CT [8,9]) and decreased bone strength (as assessed with finite element analysis [4]), despite normal BMD as shown with DXA (1). Although these data have provided valuable insights into bone microarchitecture at peripheral sites, there is no reliable technique with which to assess trabecular architecture of the lumbar spine—the site most profoundly affected by bone loss in anorexia nervosa.
CT texture analysis is an objective approach to quantifying tissue gray-level patterns by using computer-assisted measurements that are independent of subjective visual interpretation (10). Most studies have used CT texture analysis in patients with neoplasms to differentiate benign from malignant lesions, assess tumor grade, or predict survival in patients with cancer (11–15), but, to our knowledge, no studies have performed in vivo CT texture analysis of the lumbar spine to assess trabecular bone.
The purpose of our study was to determine indexes of skeletal integrity by using CT trabecular texture analysis of the lumbar spine in patients with anorexia nervosa and normal-weight control subjects and to determine clinical and body composition predictors of trabecular texture. We hypothesized that women with anorexia nervosa have CT trabecular texture that reflects impaired skeletal integrity compared with that of normal-weight control subjects and that body mass index (BMI), fat mass, and muscle mass are positively associated with skeletal integrity assessed with CT texture analysis.
Materials and Methods
Our prospective study was approved by the institutional review board and compliant with the Health Insurance Portability and Accountability Act. Data were acquired after all participants provided written informed consent before our study.
Participants
Our study was performed at a clinical research center at an academic institution. We studied a convenience sample of 60 women—30 with anorexia nervosa and 30 without anorexia nervosa and of normal weight and frequency matched for age (±2 years)—who were participants in clinical trials from November 12, 2004, to July 12, 2011, at our institution. Participants with anorexia nervosa were referred by eating disorders care providers or recruited by means of advertisements, and normal-weight control subjects were recruited by means of advertisements. Inclusion criteria for both groups were as follows: age 18–45 years and female sex. All participants were non-Hispanic and white. Participants with anorexia nervosa met weight and psychiatric criteria from the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (16). The normal-weight participants had a BMI of at least 19 kg/m2 and less than 25 kg/m2, were healthy, had regular menses, and had no history of amenorrhea or an eating disorder (Fig 1). None of the women with anorexia nervosa or normal-weight control subjects were receiving estrogen or oral contraceptives, osteoporosis medication, or other medication that could affect bone metabolism, such as glucocorticoids, gonadal steroids, or anticonvulsants. Exclusion criteria for both groups were pregnancy and presence of a chronic disease (other than anorexia nervosa). Clinical characteristics and abdominal fat cross-sectional areas have been previously reported in a subset of study participants (17–20); however, no data on trabecular texture analysis have been described.
Figure 1:
Study flowchart. QCT = quantitative CT.
CT Examinations
A 16–detector row CT scanner (LightSpeed Pro; GE Healthcare, Waukesha, Wis) was used to acquire a single section through the mid-L4 vertebral body in each participant. Participants were placed supine in the CT scanner on a calibration phantom (Mindways Software, Austin, Tex), and scanning was performed by using the following parameters: 80 kV, 70 mA, gantry rotation time of 2 seconds, 144-mm table height, and axial scanning mode. The reconstructed section thickness and field of view were kept constant at 1 cm and 48 cm, respectively. Volume CT dose index, dose-length product, and estimated effective dose were recorded.
Trabecular BMD
Analyses were performed by a postdoctoral research fellow with 2 years of experience (A.T.), who was supervised by a musculoskeletal radiologist with 11 years of experience (M.A.B.).
Trabecular BMD (in grams per cubic centimeter) of the L4 vertebral body was determined from quantitative CT by drawing a circular region of interest measuring 700 mm2 within trabecular bone; cortical bone and basivertebral veins were avoided. The mean attenuation of the vertebral body (in Hounsfield units) and the attenuation of the calibration solutions were used to calculate the mineral equivalent BMD by using the following equation: milligrams per milliliter of sample = (CT number in Hounsfield units − intercept of calibration line)/slope of calibration line.
Trabecular BMD was compared with a young female normal reference database to calculate T scores (21).
Trabecular Texture Analysis
Texture analysis was performed on the quantitative CT images by using commercially available software (TexRAD, Cambridge, England). The circular area within trabecular bone of the L4 vertebral body that was used for trabecular BMD assessment was also used for trabecular texture analysis. The following texture parameters were obtained on 12-bit images (4096 gray levels) (Fig 2): skewness (asymmetry of gray-level pixel distribution), kurtosis (pointiness or peakedness of pixel distribution), entropy (inhomogeneity of pixel distribution), and mean value of positive pixels (MPP) (the average attenuation value of pixels of greater than 0, without application of filters) (22,23).
Figure 2a:
Histogram characteristics. (a) Kurtosis is a measure of pointiness or peakedness of a distribution relative to a normal distribution. Positive kurtosis indicates a histogram that is more peaked, and negative kurtosis indicates that histogram is flatter than a normal distribution. (b) Skewness is a measure of asymmetry of a distribution. Negative skewness reflects a distribution with a tail that is longer on left, and positive skewness reflects a distribution with a tail that is longer on right.
Figure 2b:
Histogram characteristics. (a) Kurtosis is a measure of pointiness or peakedness of a distribution relative to a normal distribution. Positive kurtosis indicates a histogram that is more peaked, and negative kurtosis indicates that histogram is flatter than a normal distribution. (b) Skewness is a measure of asymmetry of a distribution. Negative skewness reflects a distribution with a tail that is longer on left, and positive skewness reflects a distribution with a tail that is longer on right.
Body Composition
Abdominal adipose tissue and paraspinal muscle compartments were quantified at the level of L4. Thresholding methods were applied to identify adipose tissue by using a threshold set for −50 to −250 HU, as described by Borkan et al (24). Manual delineation was used to separate abdominal subcutaneous adipose tissue and visceral adipose tissue. Cross-sectional areas (in square centimeters) of subcutaneous adipose tissue and visceral adipose tissue were obtained. Paraspinal muscle cross-sectional areas (in square centimeters) were determined as the sum of erector spinae, psoas major, and quadratus lumborum muscles. Analyses were performed by using software (Osirix, version 3.2.1; Pixmeo SARL, Geneva, Switzerland).
Statistical Analysis
Statistical analysis was performed by using software (JMP, version 11; SAS Institute, Cary, NC). Groups were compared by using the Student t test. All variables were tested for normality of distribution by using the Shapiro-Wilk test. Variables that were not normally distributed (age, BMI, skewness, kurtosis, visceral adipose tissue, subcutaneous adipose tissue, and paraspinal muscle cross-sectional area) were logarithmically transformed and then subjected to the t test. Pearson correlations were conducted to determine associations between different variables for normally distributed data and after logarithmic transformation of data that were not normally distributed. Multiple standard least-squares regression modeling was performed to control for BMI and age. Analyses were adjusted for multiple comparisons by using the Bonferroni method. Data are expressed as means and standard deviations. P < .05 was considered indicative of a statistically significant difference.
Results
Participant characteristics, abdominal fat and muscle cross-sectional areas, and CT trabecular texture parameters of the anorexia nervosa and normal-weight groups are shown in Table 1. Study participants ranged in age from 19 to 45 years (mean age, 27 years ± 7) and in BMI from 13.3 to 24.9 kg/m2 (mean BMI, 19.7 kg/m2 ± 3.0).
Table 1.
Clinical Characteristics, Body Composition, and Trabecular Texture of Study Participants
Note.—CSA = cross-sectional area, SAT = subcutaneous adipose tissue, VAT = visceral adipose tissue.
*Data are means ± standard deviations.
†Comparison was performed on log-transformed data.
‡Significant after controlling for multiple comparisons.
§Significant after controlling for log BMI and log age.
Duration of anorexia nervosa ranged from 5 to 324 months (mean, 142 months ± 92). Sixteen women with anorexia nervosa had amenorrhea (time of amenorrhea ranged from 4 to 144 months [mean, 41 months ± 38]). As expected, women with anorexia nervosa had lower abdominal fat and paraspinal muscle cross-sectional areas compared with normal-weight control subjects (P < .001). Radiation dose descriptors, namely volume CT dose index, dose-length product, and estimated effective doses, were 4.66 mGy, 4.66 mGy · cm, and 0.07 mSv, respectively.
All CT examinations were adequate for BMD and trabecular texture analysis. Women with anorexia nervosa had lower BMD and lower T scores than normal-weight control subjects (P < .001); however, the difference lost significance after adjustment for log BMI and log age. Women with anorexia nervosa had higher skewness and kurtosis and lower MPP compared with normal-weight control subjects (P < .001), and there was a trend toward lower entropy (P = .07) (Figs 3–5). After adjustment for log BMI and log age, the difference in all four trabecular texture parameters was significant between the groups (Table 1).
Figure 3:
Quantitative CT scan at level of L4 in 34-year-old woman of normal weight (BMI, 22 kg/m2) for assessment of trabecular BMD (139 mg/cm3).
Figure 5:
Histograms from trabecular texture analysis of patient with anorexia nervosa in Figure 4 (purple) and normal-weight control subject in Figure 3 (pink) demonstrate higher skewness, indicated by left shift of histogram toward lower CT attenuation (2.63 vs 1.63), higher kurtosis with higher peak of histogram, indicating lower variation of pixel intensities (10.33 vs 3.33), lower MPP, as indicated by fewer pixels with CT attenuation of more than 0 HU (233.1 vs 375.9), and slightly lower entropy, suggesting decreased complexity of pixel values (5.4 vs 5.7) in patient with anorexia nervosa compared with normal-weight control subject.
Figure 4:
Quantitative CT scan at level of L4 in 35-year-old woman with anorexia nervosa (BMI, 18 kg/m2) reveals trabecular BMD similar to that of patient in Figure 3 (BMD, 133 mg/cm3).
BMD was inversely associated with skewness and kurtosis and positively associated with entropy and MPP. BMI and abdominal fat and paraspinal muscle cross-sectional areas were inversely associated with skewness and kurtosis and positively associated with MPP and entropy (Table 2).
Table 2.
Correlation between CT Trabecular Texture Analysis, Body Composition, and BMD
Note.—CSA = cross-sectional area, SAT = subcutaneous adipose tissue, VAT = visceral adipose tissue.
*Significant after controlling for multiple comparisons.
Within the anorexia nervosa group, duration of amenorrhea was inversely associated with MPP (r = −0.53, P = .03) but not BMD (P = .07). Women with (n = 16) and those without (n = 14) amenorrhea showed similar BMD (P = .9). However, women with amenorrhea had higher skewness (P = .04) and higher kurtosis (P = .05) compared with women with anorexia nervosa who did not have amenorrhea; there was no difference in entropy or MPP (P = .7 and P = .9, respectively).
Lowest lifetime weight correlated positively with entropy (r = 0.45, P = .02) but not BMD (P = .1). No associations between duration of anorexia nervosa and trabecular texture parameters or BMD were observed (P > .4).
Discussion
We reported on the use of CT texture analysis for the assessment of trabecular bone in anorexia nervosa, which has not been established in the literature. We showed that trabecular texture analysis of the lumbar spine by using single-section CT is a feasible technique to determine skeletal integrity. Patients with anorexia nervosa have increased skewness and kurtosis and decreased entropy and MPP compared with normal-weight control subjects, and these parameters are strongly associated with low BMD. These findings suggest that this constellation of trabecular texture parameters reflects impaired skeletal integrity. BMI and abdominal fat and muscle cross-sectional areas were inversely associated with skewness and kurtosis and positively associated with entropy and MPP. In patients with anorexia nervosa, entropy correlated positively with lowest lifetime weight and MPP was inversely associated with duration of amenorrhea; no such associations were seen with BMD.
Anorexia nervosa is a common eating disorder that usually affects adolescents and young women, and patients with anorexia nervosa demonstrate multiple hormonal abnormalities that may alter bone turnover, leading to impaired bone accrual, bone loss, and fractures (2,3,25–27). Increasing efforts have been made to develop imaging biomarkers on the basis of DXA or CT to assess skeletal integrity and predict fracture risk in anorexia nervosa. Skeletal architecture and measures of bone strength are crucial factors for predicting fracture risk independent of BMD determination with DXA (28–31). Donaldson et al (32) used trabecular bone score, a gray-level texture parameter obtained from DXA images of the lumbar spine, in adolescents with anorexia nervosa and found positive correlations between this score and BMD obtained with DXA but no correlation with fractures. The accuracy of DXA in assessing BMD is influenced by body size (7), and phantom studies have shown that even small extraosseous soft-tissue heterogeneities can increase inaccuracies in BMD assessment (33). Therefore, the use of DXA and DXA-derived markers of trabecular integrity is limited in anorexia nervosa. Several studies have used peripheral quantitative CT, which is less influenced by body composition, to assess microarchitecture of the distal radius in adolescents and adults with anorexia nervosa. These studies revealed impaired trabecular and cortical microarchitecture and impaired bone strength at finite element analysis in patients with anorexia nervosa compared with normal-weight control subjects (1,4,8,9). Phan et al (34) used C-arm CT to assess bone microarchitecture of the lumbar spine in patients with anorexia nervosa. However, C-arm CT is limited by a small field of view and thus is feasible only in very thin patients; it is also associated with a higher radiation dose (0.7 mSv) (34) than the single-section CT used in our study (0.07 mSv).
Texture analysis is a technique that allows quantification of the spatial arrangement of pixel intensities by using a statistical evaluation of image intensities in a region of interest and can be performed with different imaging modalities. Most studies of CT texture analysis have focused on the assessment of neoplasms to determine tumor grade, treatment response, or survival. These studies have reported greater correspondence of heterogeneous tumor texture with higher grade of malignancy and lower overall treatment response and survival (11–14,35).
Some studies have performed texture analysis of bone by using radiography, magnetic resonance imaging, and ex vivo high-spatial-resolution CT. Thevenot et al (36) performed texture analysis on hip radiographs and determined texture parameters that could be used to identify patients at risk for femoral fracture. Link et al (37) performed high-spatial-resolution CT of cadaveric spine specimens with 1-mm-thick sections to assess basic texture parameters and microarchitecture. Microarchitecture was a stronger predictor of bone strength than trabecular texture (37). MacKay et al (38) assessed tibial subchondral bone in patients with osteoarthritis and found that trabecular texture, but not microarchitecture parameters, helped accurately classify patients with and patients without osteoarthritis.
We identified a combination of texture parameters assessed with single-section quantitative CT that reflect skeletal integrity in anorexia nervosa. In our study, higher skewness and kurtosis and lower entropy and MPP were associated with decreased BMD. Lower MPP reflects a lower number of pixels with CT attenuation values of more than 0, which can be seen in bone loss. Similarly, low entropy indicates decreased complexity of pixel values, which suggests impaired skeletal integrity. Higher kurtosis reflects a histogram that is narrower than a normal Gaussian distribution, indicating a lower variation of pixel intensities. Positive skewness is consistent with a histogram with a tail that is longer on the right, and our observed higher skewness in anorexia nervosa suggests that most pixels are skewed toward lower CT attenuation. In our study, trabecular texture parameters showed correlation with BMI and fat and muscle cross-sectional areas. These findings are important because anorexia nervosa is a model of malnutrition-induced bone loss. Furthermore, BMI and fat and muscle mass are known to be positively associated with BMD and bone health in this population. We found that lowest lifetime weight correlated positively with entropy and that duration of amenorrhea was inversely associated with MPP, whereas there were no such associations with BMD. In addition, women with anorexia nervosa and amenorrhea had higher skewness and kurtosis compared with women without amenorrhea, despite similar BMD. These findings suggest that trabecular texture analysis might contribute information on bone health in anorexia nervosa that is independent of BMD.
Our study had several limitations. First, the cross-sectional design limits our ability to determine causality. Second, we did not follow up patients to assess whether trabecular texture analysis can help predict fracture risk. Third, differences in body size and weight may affect trabecular texture analysis. We therefore controlled our analyses for BMI.
In conclusion, trabecular texture analysis of the lumbar spine with single-section CT is a feasible technique with which to determine skeletal integrity. Patients with anorexia nervosa had increased skewness and kurtosis and decreased entropy and MPP compared with normal-weight control subjects. These parameters were associated with lowest lifetime weight and duration of amenorrhea, whereas there were no such associations with BMD. These findings suggest that trabecular texture analysis might contribute information on bone health in anorexia nervosa that is independent of information provided by BMD. Longitudinal studies are needed to investigate whether trabecular texture analysis can improve the prediction of fracture risk in anorexia nervosa.
Advances in Knowledge
■ CT trabecular texture analysis of the lumbar spine is a feasible technique with which to determine skeletal integrity in anorexia nervosa.
■ Patients with anorexia nervosa have increased skewness and kurtosis and decreased entropy and mean value of positive pixels compared with normal-weight control subjects, which is suggestive of impaired skeletal integrity.
■ Texture parameters but not bone mineral density are associated with lowest lifetime weight and duration of amenorrhea in anorexia nervosa.
Implication for Patient Care
■ Trabecular texture analysis can be used to help assess skeletal integrity in patients with anorexia nervosa.
Received April 26, 2016; revision requested June 20; revision received July 25; accepted August 16; final version accepted August 26.
Supported by National Institutes of Health (grants K23 RR-23090, M01 RR01066-27S1, M01 RR01066, R01 MH083657, R01 DK052625, R01 HL-077674, R03 DK59297, T32 DK 007028, UL1 RR025758, 8UL1 TR000170, 1UL1 TR001102).
Disclosures of Conflicts of Interest: A.T. disclosed no relevant relationships. M.T. disclosed no relevant relationships. K.K.M. disclosed no relevant relationships. A.K. disclosed no relevant relationships. M.K.K. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: received nonfinancial support from Siemens Healthcare; received personal fees from Bracco. Other relationships: disclosed no relevant relationships. M.A.B. disclosed no relevant relationships.
Abbreviations:
- BMD
- bone mineral density
- BMI
- body mass index
- DXA
- dual x-ray absorptiometry
- MPP
- mean value of positive pixels
References
- 1.Bredella MA, Misra M, Miller KK, et al. Distal radius in adolescent girls with anorexia nervosa: trabecular structure analysis with high-resolution flat-panel volume CT. Radiology 2008;249(3):938–946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Faje AT, Fazeli PK, Miller KK, et al. Fracture risk and areal bone mineral density in adolescent females with anorexia nervosa. Int J Eat Disord 2014;47(5):458–466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Misra M, Klibanski A. Anorexia nervosa and bone. J Endocrinol 2014;221(3):R163–R176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Walsh CJ, Phan CM, Misra M, et al. Women with anorexia nervosa: finite element and trabecular structure analysis by using flat-panel volume CT. Radiology 2010;257(1):167–174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lucas AR, Melton LJ, III, Crowson CS, O’Fallon WM. Long-term fracture risk among women with anorexia nervosa: a population-based cohort study. Mayo Clin Proc 1999;74(10):972–977. [DOI] [PubMed] [Google Scholar]
- 6.Rigotti NA, Nussbaum SR, Herzog DB, Neer RM. Osteoporosis in women with anorexia nervosa. N Engl J Med 1984;311(25):1601–1606. [DOI] [PubMed] [Google Scholar]
- 7.Carter DR, Bouxsein ML, Marcus R. New approaches for interpreting projected bone densitometry data. J Bone Miner Res 1992;7(2):137–145. [DOI] [PubMed] [Google Scholar]
- 8.Faje AT, Karim L, Taylor A, et al. Adolescent girls with anorexia nervosa have impaired cortical and trabecular microarchitecture and lower estimated bone strength at the distal radius. J Clin Endocrinol Metab 2013;98(5):1923–1929. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Milos G, Spindler A, Rüegsegger P, et al. Cortical and trabecular bone density and structure in anorexia nervosa. Osteoporos Int 2005;16(7):783–790. [DOI] [PubMed] [Google Scholar]
- 10.Tourassi GD. Journey toward computer-aided diagnosis: role of image texture analysis. Radiology 1999;213(2):317–320. [DOI] [PubMed] [Google Scholar]
- 11.Ganeshan B, Abaleke S, Young RC, Chatwin CR, Miles KA. Texture analysis of non–small cell lung cancer on unenhanced computed tomography: initial evidence for a relationship with tumour glucose metabolism and stage. Cancer Imaging 2010;10:137–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ganeshan B, Goh V, Mandeville HC, Ng QS, Hoskin PJ, Miles KA. Non–small cell lung cancer: histopathologic correlates for texture parameters at CT. Radiology 2013;266(1):326–336. [DOI] [PubMed] [Google Scholar]
- 13.Ganeshan B, Miles KA, Young RC, Chatwin CR. Texture analysis in non–contrast enhanced CT: impact of malignancy on texture in apparently disease-free areas of the liver. Eur J Radiol 2009;70(1):101–110. [DOI] [PubMed] [Google Scholar]
- 14.Schieda N, Thornhill RE, Al-Subhi M, et al. Diagnosis of sarcomatoid renal cell carcinoma with CT: evaluation by qualitative imaging features and texture analysis. AJR Am J Roentgenol 2015;204(5):1013–1023. [DOI] [PubMed] [Google Scholar]
- 15.Zhang H, Graham CM, Elci O, et al. Locally advanced squamous cell carcinoma of the head and neck: CT texture and histogram analysis allow independent prediction of overall survival in patients treated with induction chemotherapy. Radiology 2013;269(3):801–809. [DOI] [PubMed] [Google Scholar]
- 16.American Psychiatric Association . Diagnostic and statistical manual of mental disorders, 5th ed. Arlington, Va: American Psychiatric Publishing, 2013. [Google Scholar]
- 17.Bredella MA, Ghomi RH, Thomas BJ, et al. Comparison of DXA and CT in the assessment of body composition in premenopausal women with obesity and anorexia nervosa. Obesity (Silver Spring) 2010;18(11):2227–2233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bredella MA, Gill CM, Keating LK, et al. Assessment of abdominal fat compartments using DXA in premenopausal women from anorexia nervosa to morbid obesity. Obesity (Silver Spring) 2013;21(12):2458–2464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gill CM, Torriani M, Murphy R, et al. Fat attenuation at CT in anorexia nervosa. Radiology 2016;279(1):151–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Miller KK, Grieco KA, Klibanski A. Testosterone administration in women with anorexia nervosa. J Clin Endocrinol Metab 2005;90(3):1428–1433. [DOI] [PubMed] [Google Scholar]
- 21.Genant HK, Cann CE, Pozzi-Mucelli RS, Kanter AS. Vertebral mineral determination by quantitative CT: clinical feasibility and normative data. J Comput Assist Tomogr 1983;7(3):554. [Google Scholar]
- 22.Miles KA, Ganeshan B, Hayball MP. CT texture analysis using the filtration-histogram method: what do the measurements mean? Cancer Imaging 2013;13(3):400–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Smith AD, Gray MR, del Campo SM, et al. Predicting overall survival in patients with metastatic melanoma on antiangiogenic therapy and RECIST stable disease on initial posttherapy images using CT texture analysis. AJR Am J Roentgenol 2015;205(3):W283–W293. [DOI] [PubMed] [Google Scholar]
- 24.Borkan GA, Gerzof SG, Robbins AH, Hults DE, Silbert CK, Silbert JE. Assessment of abdominal fat content by computed tomography. Am J Clin Nutr 1982;36(1):172–177. [DOI] [PubMed] [Google Scholar]
- 25.Gordon CM, Goodman E, Emans SJ, et al. Physiologic regulators of bone turnover in young women with anorexia nervosa. J Pediatr 2002;141(1):64–70. [DOI] [PubMed] [Google Scholar]
- 26.Misra M, Klibanski A. Endocrine consequences of anorexia nervosa. Lancet Diabetes Endocrinol 2014;2(7):581–592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Misra M, Klibanski A. Anorexia nervosa and its associated endocrinopathy in young people. Horm Res Paediatr 2016;85(3):147–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab 2005;90(12):6508–6515. [DOI] [PubMed] [Google Scholar]
- 29.Engelke K, Lang T, Khosla S, et al. Clinical use of quantitative computed tomography–based advanced techniques in the management of osteoporosis in adults: the 2015 ISCD official positions—part III. J Clin Densitom 2015;18(3):393–407. [DOI] [PubMed] [Google Scholar]
- 30.Engelke K, Libanati C, Fuerst T, Zysset P, Genant HK. Advanced CT based in vivo methods for the assessment of bone density, structure, and strength. Curr Osteoporos Rep 2013;11(3):246–255. [DOI] [PubMed] [Google Scholar]
- 31.Zysset P, Qin L, Lang T, et al. Clinical use of quantitative computed tomography–based finite element analysis of the hip and spine in the management of osteoporosis in adults: the 2015 ISCD official positions—part II. J Clin Densitom 2015;18(3):359–392. [DOI] [PubMed] [Google Scholar]
- 32.Donaldson AA, Feldman HA, O’Donnell JM, Gopalakrishnan G, Gordon CM. Spinal bone texture assessed by trabecular bone score in adolescent girls with anorexia nervosa. J Clin Endocrinol Metab 2015;100(9):3436–3442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bolotin HH, Sievänen H, Grashuis JL. Patient-specific DXA bone mineral density inaccuracies: quantitative effects of nonuniform extraosseous fat distributions. J Bone Miner Res 2003;18(6):1020–1027. [DOI] [PubMed] [Google Scholar]
- 34.Phan CM, Khalilzadeh O, Dinkel J, et al. C-arm CT for histomorphometric evaluation of lumbar spine trabecular microarchitecture: a study on anorexia nervosa patients. Br J Radiol 2013;86(1027):20120451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ravanelli M, Farina D, Morassi M, et al. Texture analysis of advanced non–small cell lung cancer (NSCLC) on contrast-enhanced computed tomography: prediction of the response to the first-line chemotherapy. Eur Radiol 2013;23(12):3450–3455. [DOI] [PubMed] [Google Scholar]
- 36.Thevenot J, Hirvasniemi J, Pulkkinen P, et al. Assessment of risk of femoral neck fracture with radiographic texture parameters: a retrospective study. Radiology 2014;272(1):184–191. [DOI] [PubMed] [Google Scholar]
- 37.Link TM, Majumdar S, Lin JC, et al. Assessment of trabecular structure using high resolution CT images and texture analysis. J Comput Assist Tomogr 1998;22(1):15–24. [DOI] [PubMed] [Google Scholar]
- 38.MacKay JW, Murray PJ, Low SB, et al. Quantitative analysis of tibial subchondral bone: texture analysis outperforms conventional trabecular microarchitecture analysis. J Magn Reson Imaging 2016;43(5):1159–1170. [DOI] [PubMed] [Google Scholar]