Single-slice MRI for body composition assessment: repeatability, reproducibility, and observer variability

doi:10.21203/rs.3.rs-5260409/v1

Purpose

The single-slice MRI at L3 vertebra offers an efficient way to assess body composition but the evidence on the reliability of this method is scarce. This study evaluates the accuracy and precision of this method for adipose and muscle tissue measurements.

Methods

The technical performance of single-slice (L3) MRI body composition measurements was assessed in a prospective study of 12 participants, focusing on scan-rescan repeatability, cross-scanner reproducibility, and analyst variability. Additionally, retrospective data from 36 participants were analyzed to evaluate inter-device and inter-observer (analyst vs. radiologist) variability across a wide range of scanners and body types. Blinded analyses were performed for visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) cross-sectional areas (CSA), VAT and SAT indices, VAT/SAT ratio, skeletal muscle CSA, skeletal muscle index (SMI), psoas muscle CSA, and psoas muscle index.

Results

Single-slice MRI-based body composition measurements showed high scan-rescan repeatability (CoV: 1.5%–7.9%, ICC: 0.97–1.0), with low repeatability coefficients (RC) across key metrics, including 12 cm² for SAT CSA, 15 cm² for VAT CSA, 5 cm² for skeletal muscle CSA, and 1.4 cm² for psoas muscle CSA. Cross-scanner reproducibility was consistent (CoV: 2.3%–15%, ICC: 0.90–1.0). Inter- and intra-analyst variability was minimal (CoV: 0.5%–5.0%, ICC: 0.98–1.0). Analyst-radiologist comparisons showed near-perfect correlations (r = 0.97–1.00, p < 0.001) and excellent reliability (ICC: 0.96–1.0).

Conclusion

The results demonstrate that MRI-based single-slice method at the L3 vertebral level provides accurate, repeatable, and reproducible measurements of adipose and muscle tissue across a wide range of body types, consistent between trained analysts and experienced radiologists. These findings support the method’s accuracy and consistency for longitudinal assessments.

2D body composition

magnetic resonance imaging

repeatability

reproducibility

L3

Obesity and sarcopenia are significant public health issues with a rising prevalence globally, both contributing to increased healthcare costs and reduced quality of life [1–4]. Obesity increases the risk of chronic diseases such as cardiovascular disease [5], chronic kidney disease [6, 7], diabetes [8], metabolic dysfunction-associated liver disease (MASLD) [9, 10], and certain types of cancer [11, 12]. Similarly, sarcopenia, defined as the loss of skeletal muscle mass and strength, is associated with increased adverse outcomes including falls, functional decline, frailty, diabetes, and mortality [13–16]. Anthropometric measures such as body mass index (BMI), waist circumference, and waist-to-hip ratio are widely used to estimate fat distribution but they are inadequate for accurately predicting individual fat distribution and metabolic risk [17, 18]. BMI, in particular, is a crude measure that fails to differentiate between fat and muscle mass, both of which are crucial for assessing metabolic health. This limitation is particularly evident in conditions like sarcopenia and cachexia, where significant muscle wasting occurs. In sarcopenia, age-related muscle loss may be accompanied by increased adiposity, which can result in a ‘normal’ BMI despite significant muscle loss. Cachexia, associated with chronic illness, involves rapid muscle and weight loss, affecting BMI differently, yet failing to fully capture the underlying changes in body composition. Therefore, there is a critical need for advanced body composition analysis tools that can provide a more accurate assessment of fat distribution and muscle mass. Such tools are essential not only for the effective evaluation and management of conditions such as obesity and sarcopenia but also hold significant value in clinical trials for monitoring the efficacy of various weight loss interventions, including pharmacological agents [19, 20], bariatric surgery [20, 21], and lifestyle modifications [20, 22]. Furthermore, these tools are useful in predicting clinical outcomes beyond mere weight loss measures. They can provide important information about a patient's potential response to major interventions such as surgery [23] or chemotherapy [24], allowing for more personalized and effective treatment planning. This predictive capability enhances risk assessment and can guide clinical decision-making, potentially improving patient outcomes across various medical disciplines [25].

Whole-body magnetic resonance imaging (MRI) is the gold standard for assessing the skeletal muscle and adipose tissue volumes [26]. It has been validated to reliably and accurately measure body compartments [27–29], and its lack of ionizing radiation, combined with its applicability to nearly all cohorts and ages, makes it suitable for repeated scans in longitudinal studies and for monitoring interventions [26]. However, its application is often restricted due to high costs, time-consuming manual post-processing, and the increasing demands on an already overburdened radiology workforce, further hampering its widespread use. Alternatives to whole-body solutions have focused on measurements from single- or few-slice segmentations. Previous research has shown that single-slice measurements at the level of third lumbar vertebra (L3) are optimal in terms of reproducibility and correlations with whole-body compartments [30, 31], thereby simplifying data acquisition, reducing assessment time, complexity and overall costs. However, there is a lack of studies reporting on the technical performance and reliability of these measurements. Detailed technical validation is crucial for broader adoption of such methodology in routine clinical workflows and clinical trials. The present study aims to assess the accuracy and precision of MRI-based single-slice L3 measurements of adipose and muscle tissue.

Study design and population

The evaluation of the technical performance of single-slice body composition measurements was based on both prospective and retrospective cohorts. All studies received appropriate approvals from the West of Scotland Research Ethics Committee (REC reference: 21/WS/0066) and were performed in accordance with the principles of the Declaration of Helsinki. All participants provided informed consent to participate, including those who had previously taken part in ethically approved studies and consented to the use of their data.

Two primary software tools were used in this study for body composition analysis. The first tool, OsiriX MD (Pixmeo, Switzerland), is an FDA-cleared, commercially available medical imaging software used as the reference standard in this study. It provides general-purpose tools for medical image analysis, including both manual and semi-automated segmentation capabilities. This will hereafter be referred to as the "reference software." The second tool, VitruvianScan™ (Perspectum Ltd., UK), is a specialized single-slice (2D) body composition MRI analysis software. It incorporates semi-automated segmentation tools specifically designed for adipose and muscle tissue quantification at the L3 vertebral level. This will hereafter be referred to as the "study software." Both software packages were used to analyze the same MRI datasets, allowing for comparison of their performance in body composition assessment.

In the prospective study, 12 participants were recruited. Participants were scanned on Philips 1.5T (reference scanner) and Philips 3T scanners (Fig. 1A). For scan-rescan repeatability, each participant was scanned twice under the same conditions, with the patient leaving the MRI table between scans and being re-localized and re-registered with a new blinding code for each scan, with a maximum 30-minute interval between the scans. Cross-scanner reproducibility involved scanning each participant on the same day with both scanners, with a maximum 1-hour interval between the scans (Fig. 1A). Inter- and intra-analyst variability assessments were conducted on data acquired from the Philips 1.5T (reference) scanner. Adult subjects aged 18 years or older, with BMI between 20 and 35 kg/m², and with a range of body types (from muscular to frail) were included. Exclusion criteria comprised the presence of any MRI contraindications (e.g. pacemaker, shrapnel injury, severe claustrophobia) [32].

In addition, retrospective data from 36 participants were selected to ensure comprehensive representation across a wide range of body types, scanner manufacturers (Siemens, Philips, GE), and field strengths (1.5T and 3T) (Fig. 1B). To achieve balanced representation, six participants were chosen for each combination of scanner manufacturer and field strength, with equal distribution across sex (3 females, 3 males) and BMI categories (lean, normal weight, obese). This dataset was used for inter-device assessments, comparing the performance of the study software with the reference software. Additionally, inter-reader assessment was performed by having trained analysts and radiologists independently analyze the same images using the study software to compare performance across different reader groups (Fig. 1B). Inclusion criteria ensured a diverse dataset, covering a range of body types (BMI, sex), scanner models (Siemens, Philips, GE at both 1.5T and 3T), and known height for all participants.

Data acquisition

In the prospective testing, all individuals were scanned using Philips Ingenia 1.5T and Philips dStream 3T. Data acquisitions were performed at the Leeds General Infirmary (The Leeds Teaching Hospitals NHS Trust), United Kingdom.

The MRI data were acquired using a 3D SPGR (Spoiled Gradient Echo) sequence with the Dixon option in the axial plane, a flip angle of 10° (1.5T) or 6° (3T), and minimum repetition time (TR). The maximum field of view (FOV) was set in both the right-left (RL) and anterior-posterior (AP) directions. Imaging was performed with an in-plane resolution of 2 × 2 mm² and a slice thickness of 6 mm, with a reconstructed slice thickness set to 3 mm (interpolation on). Parallel imaging techniques were utilized with acceleration factors of 2 in the phase-encoding direction and 2 in the slice-encoding direction with SENSE. A single average was used for acquisitions at 1.5T, while two averages were used for acquisitions at 3T. The number of slices was adjusted to achieve an acquisition time of approximately 12 seconds, with data collected in a single breath-hold and centred on the L3 vertebra.

The retrospective MRI data had been acquired using a 3D SPGR Dixon sequence. The in-plane resolution was between 1 x 1 mm² and 4 x 4 mm². The acquired slice thickness was between 1 mm and 10 mm. The flip angle was required to be less than or equal to 16° and 12° at 1.5T and 3T, respectively.

Data processing

Anonymised MR data were analyzed using both the study software and the reference software, as described earlier. Throughout all analyses, the radiologists and analysts remained blinded to the subject identities, scanner manufacturers/models, field strengths, scan session details, and any analyses or results produced by other radiologists or analysts. The analyses were performed by trained analysts with a minimum of 2 years of experience in MRI quantitative analysis and/or radiologists with a minimum of 5 years of experience in abdominal MRI. Analysts and radiologists underwent comprehensive training on the study software and had prior experience performing single-slice analyses at the L3 vertebral level.

Body composition analysis involved manual segmentations on a single L3 slice to obtain cross-sectional area (CSA) measurements (in cm²) for four classes; the visceral adipose tissue (VAT), the subcutaneous adipose tissue (SAT), the skeletal and the psoas muscles (Fig. 2).

The assessment of observer variability included: prospective intra- and inter-analyst evaluations as well as a retrospective inter-reader comparison (analysts vs. radiologists). For the prospective intra-analyst variability assessment, the same analyst analyzed a sub-sample of scans from the reference scanner twice (analysis 1 and analysis 2), presented in a random order and under blinded conditions, with a maximum of 4 days between the reads. Three analysts processed the same set of cases twice, with the anonymized cases presented randomly each time. For inter-analyst assessment, three analysts analyzed the same sub-sample of scans from the reference scanner under blinded conditions. For the retrospective inter-reader variability assessment (analysts vs. radiologists), a “gold standard” measurement was created on the median of successful measurements by the three radiologists using the study software. This “gold standard” was then used to evaluate the performance of each of the analysts.

The retrospective inter-device assessment compared results from the study software and the reference software on the same set of images. Three radiologists used the study software, while one of these radiologists (designated as the reference radiologist) used the reference software. The median results from the three radiologists using the study software were compared with the results from the reference radiologist using the reference software. To generate results using the reference software, the reference radiologist was presented with a single axial slice at the center of the L3 vertebrae. Initially, a region-growing tool within the software, utilizing image-dependent threshold values, was used to generate coarse masks for the four tissue classes. Subsequently, the reference radiologist edited these masks using a brush tool, where necessary, to generate the final output masks. The regions of interest (ROIs) were then exported, and the CSA measurements parsed from the resulting output file.

Within the study software, pre-processing was performed to generate a signal fat fraction (sFF) map calculated on a voxel-wise basis:

$$\:sFF=\frac{F}{F+W}$$

where F represents the signal in the fat image generated from the 3D SPGR acquisition, and W represents the signal in the corresponding water image. Analysts were presented with sagittal views of the fat, water, and calculated sFF maps, with a line indicating the central slice (Fig. 2a). This central slice was identical to the slice used by the radiologist in the reference device. Upon loading the central slice, the analysts used a circular brush tool to segment regions corresponding to a specific tissue class. During this interactive process, as brush strokes were applied to the image to generate an output segmentation mask, voxels within the bounds of the brush were only added to the segmentation mask if specific criteria were met in the intensity values of the corresponding voxels in the sFF image (Fig. 3). For VAT and SAT classes, voxels were included where sFF ≥ 0.5. For skeletal muscle and psoas classes, voxels were included where sFF < 0.5.

After analyses were completed by analysts and radiologists and measurements exported, these data were subsequently evaluated using an automated statistical pipeline.

Statistical analysis

For the prospective testing of repeatability, reproducibility, and intra/inter-analyst variability, the following statistical analyses were performed: Bland Altman analysis (bias and 95% limits of agreement, LoA) [33, 34], standard deviation (SD) of differences, mean coefficient of variation (CoV, in %), intra-class coefficient (ICC, two-way mixed effects, absolute agreement, single measurement), and repeatability coefficient (RC) or reproducibility coefficient (RDC). The RC/RDC, an estimate of the maximum difference that only 5% of measurement pairs will exceed asymptotically, was calculated according to standard formulae [35, 36]. Agreement was classified as follows: ICC > 0.9, excellent; 0.9–0.75, good; 0.75–0.5, moderate; < 0.5, poor [37]. These statistical analyses were also performed for the retrospective evaluation of inter- and inter-reader (analyst vs. radiologist) variability. In addition, Pearson correlation coefficients and corresponding p-values were calculated for inter-reader assessments. The statistical analyses were performed for all CSA (cm²): VAT, SAT, skeletal muscle, and psoas muscle; VAT/SAT ratio (no units); and VAT, SAT, skeletal muscle, and psoas indices (cm²/m²). Inter-device variability assessments were conducted only for CSAs. The indices were calculated by indexing the body composition CSA measurements (cm²) to height by dividing it by the height squared (m²). These height-standardized measurements can be used for comparison between individuals of different height.

In the analysis of participant demographics, values are presented as mean ± standard deviation (SD) for normally distributed data and as median with interquartile range (IQR) for non-normally distributed data. Descriptive statistics were also calculated for all prospective body composition measurements with all study subjects pooled.

The required sample size for the performance testing of the MRI body composition measurements was estimated based on Bland Altman estimates using previously described methodology [38].

All statistical analyses were performed using scripts written in the Python (v3.10).

Participant characteristics

In the prospective data analysis, 12 participants were included (5 females), with a mean age of 31 (± 10) years and a mean BMI of 23.6 (± 4.3) kg/m². Additionally, 36 individuals (18 females) were included in the retrospective inter-device and inter-reader analysis, with a median age of 31 (24–43) years and a median BMI of 24.9 (21.5–28.0) kg/m². Individual subject demographics are reported in Supplementary Tables S1 and S2.

Median values for single-slice body composition measurements

Descriptive statistics (median and quartiles of each measurement variable) for the prospective study cohort (n = 12) are presented in Table 1.

Table 1. Descriptive statistics of the prospective study cohort

Measurement	Median	Q1	Q3
Psoas muscle CSA (cm²)	23.5	16.4	34.3
Psoas muscle Index (cm²/m²)	7.7	5.7	10.4
Skeletal muscle CSA (cm²)	125.5	106.0	174.5
Skeletal muscle Index (cm²/m²)	41.1	38.0	50.8
SAT CSA (cm²)	124.0	101.0	178.5
SAT Index (cm²/m²)	44.1	30.7	61.0
VAT CSA (cm²)	65.9	47.4	93.6
VAT Index (cm²/m²)	25.4	14.6	38.3
VAT/SAT Ratio	0.49	0.29	0.80

Scan-rescan repeatability

The scan-rescan repeatability results (bias, LoA, SD, ICC, RC, and CoV) for each body composition measurement for the reference scanner (Philips 1.5T) are reported in Table 2. The RC values were: 1.4 cm² for psoas muscle CSA, 0.44 cm²/m²for psoas muscle index, 5.1 cm² for skeletal muscle CSA, 1.9 cm²/m² for skeletal muscle index, 12 cm²for SAT CSA, 3.5 cm²/m² for SAT index, 15 cm² for VAT CSA, 4.5 cm²/m²for VAT index, and 0.099 for VAT/SAT ratio. All measurements showed relatively low variability (CoV 1.5% to 7.9%) and excellent reliability (ICC 0.98 to 1.0) across the tested scanner field strengths (Table 2 and Supplementary Table S3). Overall, VAT-related measurements consistently showed greater variability compared to measurements across all testing conditions (CoV 5.0% to 7.9%).

Cross-scanner reproducibility

The reproducibility results (bias, LoA, SD, ICC, RC, and CoV) for the cross-scanner comparisons between the Philips 1.5T and Philips 3T are reported in Table 3. The RDC values were: 4.6 cm² for psoas muscle CSA, 1.6 cm²/m²for psoas muscle index, 7.5 cm²for skeletal muscle CSA, 2.8 cm²/m² for skeletal muscle index, 26 cm²for SAT CSA, 8.4 cm²/m²for SAT index, 42 cm² for VAT CSA, 12 cm²/m² for VAT index, and 0.27 for VAT/SAT ratio. All measurements showed low to moderate variability (CoV 2.3% to 15%) and excellent reliability (ICC 0.90 to 1.0).

Table 2. Summary of scan-rescan repeatability results for body composition measurements for the reference scanner (Philips 1.5T).

Measurement	Bias	Lower LoA	Upper LoA	SD	ICC	RC	CoV
Psoas muscle CSA (cm²)	-0.44	-1.5	0.65	0.56	1.0	1.4	2.0
Psoas muscle Index (cm²/m²)	-0.15	-0.49	0.19	0.17	1.0	0.44	2.0
Skeletal muscle CSA (cm²)	0.12	-5.2	5.4	2.7	1.0	5.1	1.5
Skeletal muscle Index (cm²/m²)	0.050	-2.0	2.1	1.0	0.99	1.9	1.7
SAT CSA (cm²)	-1.8	-13	9.7	5.9	1.0	12	1.9
SAT Index (cm²/m²)	-0.52	-4.0	3.0	1.8	1.0	3.5	1.9
VAT CSA (cm²)	1.7	-14	17	7.9	0.99	15	5.0
VAT Index (cm²/m²)	0.38	-4.3	5.0	2.4	0.98	4.5	5.0
VAT/SAT Ratio	0.012	-0.089	0.11	0.051	0.98	0.099	5.9

Abbreviations: CoV, coefficient of variation; CSA, cross-sectional area, ICC, intraclass coefficient; LoA, limits of agreement; RC, repeatability coefficient; SAT, subcutaneous adipose tissue; SD, standard deviation; VAT, visceral adipose tissue

Table 3. Summary of cross-scanner reproducibility results for body composition measurements (Philips 3T vs. Philips 1.5T reference scanner).

Measurement	Bias	Lower LoA	Upper LoA	SD	ICC	RDC	CoV
Psoas muscle CSA (cm²)	-0.3	-5.1	4.5	2.5	0.97	4.6	5.3
Psoas muscle Index (cm²/m²)	-0.085	-1.7	1.5	0.83	0.95	1.6	5.3
Skeletal muscle CSA (cm²)	1.9	-4.9	8.7	3.5	1.00	7.5	2.3
Skeletal muscle Index (cm²/m²)	0.71	-1.8	3.2	1.3	0.99	2.8	2.4
SAT CSA (cm²)	-7.3	-30	15	11	0.99	26	7.3
SAT Index (cm²/m²)	-2.3	-9.6	5.0	3.7	0.99	8.4	7.5
VAT CSA (cm²)	8.6	-31	49	20	0.92	42	10
VAT Index (cm²/m²)	2.4	-8.8	14	5.7	0.90	12	10
VAT/SAT Ratio	0.076	-0.16	0.31	0.12	0.90	0.27	15

Abbreviations: CoV, coefficient of variation; ICC, intraclass coefficient; LoA, limits of agreement; RC, repeatability coefficient; SAT, subcutaneous adipose tissue; SD, standard deviation; VAT, visceral adipose tissue

Inter- and intra-analyst variability

Both inter- and intra-analyst variability measures demonstrated excellent agreement for all body composition measurements, with ICC values ranging from 0.98 to 1.0, and low CoV values (0.5% to 5.0%). Detailed performance metrics are reported in Supplementary Table S4 for inter-analyst variability and in Supplementary Table S5 for intra-analyst variability for all three analysts.

Inter-reader (analysts vs. radiologists) variability

Each trained analyst using the study software demonstrated performance equivalent to the gold standard, as defined by the median results of three expert radiologists using the device. Pearson correlation coefficients indicated near-perfect correlation across all measurements, scanner field strengths, and manufacturers (r values: 0.97 to 1.00, p < 0.001). The ICC values were excellent, ranging from 0.94 to 1.00 (Supplementary Table S6).

Inter-device variability

The performance of the study software, as assessed by three radiologists, was found to be largely equivalent to that of the radiologist using the reference software for the tested body composition measurements: VAT CSA, SAT CSA, psoas muscle CSA, skeletal muscle CSA (Supplementary Table S7). The reliability was generally good to excellent (ICC 0.84 to 1.0), except for skeletal muscle CSA at Siemens 1.5T and Philips 3T, and psoas muscle CSA at GE 3T, where the reliability was moderate (ICC 0.60, 0.67, 0.65, respectively). Upon further examination, measurement discrepancies were identified in 4 (out of 36) cases. The differences were explored in detail through separate interviews with each radiologist, who independently scored the segmentation masks generated by the study software and the regulated reference software. The consensus among the radiologists was that the study software provided measurements that were either comparable to or superior to those from the reference software.

In this study, we have demonstrated that single-slice MRI body composition measurements at the L3 vertebra level provide consistent and reliable results across a diverse cohort, including both lean and obese individuals. All measurements showed high scan-rescan repeatability and cross-scanner reproducibility, with minimal inter/intra-observer and inter-device variability.

Single-slice measurements at the L3 vertebra have been shown to [31] correlate strongly with volumetric assessments of muscle and adipose tissue, offering reliable insights comparable to those derived from full volumetric techniques [30,31,39]. While previous technical evaluations have focused on volumetric assessments [40–42] or single-slice assessments of paraspinal muscles [43,44], our study provides a comprehensive investigation into the technical validity of single-slice body composition measurements of both muscle and adipose tissue.

The scan-rescan repeatability is a valuable for longitudinal studies, where consistent measurements enable accurate monitoring of disease progression and treatment effects. We assessed scan-rescan repeatability using the coefficient of variation (CoV), which standardizes the spread of differences between repeated measurements relative to the mean. This approach facilitates comparison of test-retest repeatability across different body composition compartments and measurement units, providing a robust evaluation of the method's reliability for clinical applications. Our results demonstrate low CoV for scan-rescan repeatability across all single-slice measurements, (1.5% to 7.9%). These results align with previous studies, which reported comparable CoV values (1% to 13%) for volumetric assessments of adipose tissue, including SAT, VAT, and total adipose tissue, with higher variability noted for manually segmented metrics [41,45–48]. CoV values were consistently higher than those for intra- and inter-analyst variability for skeletal muscle CSA, SAT CSA, and VAT CSA, indicating that acquisition-related factors, such as thermal noise, scanner calibration, and inherent technical variations, were the primary contributors to variance. Our consecutive MRI scan design minimized physiological changes, further emphasizing the impact of these acquisition-related factors. Conversely, for psoas muscle CSA, the CoV for scan-rescan repeatability was consistently lower than the CoV for intra- and inter-analyst assessments. This suggests that analyst-related factors contribute more to the variance observed for this smaller structure, possibly due to its susceptibility to minor discrepancies in measurement techniques and interpretation.

The repeatability coefficient (RC), representing the smallest detectable difference with 95% confidence, provides context for the clinical applicability of our measurements. Our RC values (15 cm² for VAT and 12 cm² for SAT) are much lower than the changes reported in a recent meta-analysis of weight-loss drug interventions (31.6 cm² for VAT and 43.0 cm² for SAT) [20], indicating sufficient sensitivity for detecting clinically relevant changes, particularly in populations with elevated baseline adiposity.

Reproducibility across different scanners is crucial for multi-centre studies and standardized clinical practice. Our study found higher cross-scanner reproducibility CoV for SAT (7.3%) and VAT (10%) CSA compared to previous volumetric measurements (3.4% and 4.4%, respectively) [39]. This difference likely reflects the greater susceptibility of single-slice measurements to inter-scanner variations. However, the L3 single-slice method still offers reliable reproducibility across different scanners. Notably, the reproducibility coefficient exceeded the repeatability coefficient for all measurements, indicating that inter-scanner variability is the primary source of measurement inconsistency. These findings suggest that while the L3 single-slice method is sufficiently sensitive and reproducible for clinical applications, careful consideration of inter-scanner variability is necessary when designing multi-centre studies or comparing results across different imaging systems.

All body composition measurements performed here exhibited high consistency in inter- and intra-analyst variability assessments, as well as inter-reader (analysts vs. radiologists) assessments, as tested across various scanner field strengths and manufacturers. The inter- and intra-observer agreement was excellent (ICC 0.98-1.0) and similar to the previously reported values for skeletal and paraspinal muscle CSA (ICC 0.95-1.0) [43,49], and VAT volume (ICC 0.99) [41]. Additionally, the intra-observer variability was comparable to the inter-observer variability. This suggests that the method is less user-dependent, allowing for interchangeable use of operators in longitudinal studies.

The single-slice MRI method offers an efficient alternative to volumetric techniques, with shorter acquisition (<1 min) and processing times (~5 min). Its consistent scan-rescan reliability and cross-scanner reproducibility could potentially reduce patient numbers and shorten clinical trial durations, leading to cost saving. Additionally, in this study, multiple scans were successfully acquired for each participant without any acquisition failure in any session, a level of reliability typically not achieved with full volumetric approaches.

There were several limitations in this study. The analyses relied on manual segmentation techniques. Automated or semi-automated approaches could potentially improve reliability, particularly for complex measurements, such as VAT, where the likelihood of measurement error is higher due to intricate anatomy and variability of this tissue. In addition, our study included participants with a BMI range of 18-31 kg/m², covering normal weight, overweight, and mildly obese individuals, but did not include those with severe obesity (BMI > 35 kg/m²). Interestingly, a recent study demonstrated that the reliability of total adipose tissue measurements, for example, was maintained in obese patients [41]. Future research should explore whether the RC for single-slice measurements remains consistent with higher adiposity levels.

In conclusion, this study demonstrates the reliability of single-slice body composition measurements across multiple MRI scanner field strengths and manufacturers. The method's efficiency, reproducibility, and low observer variability support its utility in disease assessment and drug effect monitoring across various clinical settings, contributing to the standardization of body composition MRI procedures.

Financial interests

MN, LN, CEH, TD, LFC, RS, HTB, and MDR are employees at Perspectum Ltd, the company that developed VitruvianScan. HTB and MDR are also shareholders at Perspectum Ltd. MP is a consultant for and a shareholder at Perspectum.

Competing Interests

MN, LN, CEH, TD, LFC, RS, HTB, and MDR are employees at Perspectum Ltd, the company that developed VitruvianScan. HTB and MDR are also shareholders at Perspectum Ltd. MP is a consultant for and a shareholder at Perspectum.

Funding

The study was funded by Perspectum Ltd.

Author Contribution

MDR, MP, CEH, HTB, and RS contributed to the study conception and design. Material preparation, data collection and analysis were performed by LN, TD, CEH, LFP, MN, RS, MP, GS, and AM. The first draft of the manuscript was written by MN. Review and editing were performed by MDR, MP, HTB, CEH, LN, and TD. All authors read and approved the final manuscript.

Data Availability

Data supporting the findings of this study are included within the manuscript and supplementary information. Additional data may be made available upon reasonable request to qualified researchers, subject to legal and regulatory requirements.

Boutari C, Mantzoros CS. A 2022 update on the epidemiology of obesity and a call to action: as its twin COVID-19 pandemic appears to be receding, the obesity and dysmetabolism pandemic continues to rage on. Metabolism 133, 155217 (2022).
Cawley J, Biener A, Meyerhoefer C et al. Direct medical costs of obesity in the United States and the most populous states. J Manag Care Spec Pharm 27(3), 10.18553/jmcp.2021.20410 (2021).
Petermann‐Rocha F, Balntzi V, Gray SR et al. Global prevalence of sarcopenia and severe sarcopenia: a systematic review and meta‐analysis. J Cachexia Sarcopenia Muscle 13(1), 86–99 (2022).
Dennison EM, Sayer AA, Cooper C. Epidemiology of sarcopenia and insight into possible therapeutic targets. Nat Rev Rheumatol 13(6), 340–347 (2017).
Emerging Risk Factors Collaboration, Wormser D, Kaptoge S et al. Separate and combined associations of body-mass index and abdominal adiposity with cardiovascular disease: collaborative analysis of 58 prospective studies. Lancet 377(9771), 1085–1095 (2011).
Kovesdy CP, Furth SL, Zoccali C. Obesity and Kidney Disease. Can J Kidney Health Dis 4, 2054358117698669 (2017).
Nawaz S, Chinnadurai R, Al‐Chalabi S et al. Obesity and chronic kidney disease: A current review. Obes Sci Pract 9(2), 61–74 (2022).
Singh GM, Danaei G, Farzadfar F et al. The age-specific quantitative effects of metabolic risk factors on cardiovascular diseases and diabetes: a pooled analysis. PLoS One 8(7), e65174 (2013).
Karlas T, Wiegand J, Berg T. Gastrointestinal complications of obesity: Non-alcoholic fatty liver disease (NAFLD) and its sequelae. Best Practice & Research Clinical Endocrinology & Metabolism 27(2), 195–208 (2013).
Huttasch M, Roden M, Kahl S. Obesity and MASLD: Is weight loss the (only) key to treat metabolic liver disease? Metabolism 157, 155937 (2024).
Pati S, Irfan W, Jameel A, Ahmed S, Shahid RK. Obesity and Cancer: A Current Overview of Epidemiology, Pathogenesis, Outcomes, and Management. Cancers (Basel) 15(2), 485 (2023).
Rathmell JC. Obesity, Immunity, and Cancer. New England Journal of Medicine 384(12), 1160–1162 (2021).
Chen H, Huang X, Dong M, Wen S, Zhou L, Yuan X. The Association Between Sarcopenia and Diabetes: From Pathophysiology Mechanism to Therapeutic Strategy. Diabetes Metab Syndr Obes 16, 1541–1554 (2023).
Mesinovic J, Zengin A, Courten BD, Ebeling PR, Scott D. Sarcopenia and type 2 diabetes mellitus: a bidirectional relationship. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy 12, 1057 (2019).
Cruz-Jentoft AJ, Sayer AA. Sarcopenia. The Lancet 393(10191), 2636–2646 (2019).
Park B, Bhat S, Xia W et al. Consensus-defined sarcopenia predicts adverse outcomes after elective abdominal surgery: meta-analysis. BJS Open 7(4), zrad065 (2023).
Thomas EL, Frost G, Taylor-Robinson SD, Bell JD. Excess body fat in obese and normal-weight subjects. Nutr Res Rev 25(1), 150–161 (2012).
Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2(3), 141–147 (2001).
Conte C, Hall KD, Klein S. Is Weight Loss–Induced Muscle Mass Loss Clinically Relevant? JAMA 332(1), 9–10 (2024).
Merlotti C, Ceriani V, Morabito A, Pontiroli AE. Subcutaneous fat loss is greater than visceral fat loss with diet and exercise, weight-loss promoting drugs and bariatric surgery: a critical review and meta-analysis. Int J Obes 41(5), 672–682 (2017).
Meyer-Gerspach AC, Peterli R, Moor M et al. Quantification of Liver, Subcutaneous, and Visceral Adipose Tissues by MRI Before and After Bariatric Surgery. OBES SURG 29(9), 2795–2805 (2019).
Ross R, Rissanen J, Pedwell H, Clifford J, Shragge P. Influence of diet and exercise on skeletal muscle and visceral adipose tissue in men. J Appl Physiol (1985) 81(6), 2445–2455 (1996).
Hua X, Deng J-P, Long Z-Q et al. Prognostic significance of the skeletal muscle index and an inflammation biomarker in patients with breast cancer who underwent postoperative adjuvant radiotherapy. Current Problems in Cancer 44(2), 100513 (2020).
Prado CM, Lieffers JR, McCargar LJ et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. The Lancet Oncology 9(7), 629–635 (2008).
Donato B, Almeida R, Raimundo M et al. Myosteatosis: an underrecognized risk factor for mortality in non-dialysis chronic kidney disease patients. J Nephrol (2024).
Heymsfield SB, Smith B, Robson MD. Measurement of Total Adiposity, Regional Fat Depots, and Ectopic Fat. In: Handbook of Obesity - Volume 1 (4th Edition). CRC Press, (2023).
Mitra S, Fernandez-Del-Valle M, Hill JE. The role of MRI in understanding the underlying mechanisms in obesity associated diseases. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease 1863(5), 1115–1131 (2017).
Nordez A, Jolivet E, Südhoff I, Bonneau D, de Guise JA, Skalli W. Comparison of methods to assess quadriceps muscle volume using magnetic resonance imaging. J Magn Reson Imaging 30(5), 1116–1123 (2009).
Selberg O, Burchert W, Graubner G, Wenner C, Ehrenheim C, Müller MJ. Determination of anatomical skeletal muscle mass by whole body nuclear magnetic resonance. Basic Life Sci 60, 95–97 (1993).
Maislin G, Ahmed MM, Gooneratne N et al. Single Slice vs. Volumetric MR Assessment of Visceral Adipose Tissue - Reliability and Validity Among the Overweight and Obese. Obesity (Silver Spring) 20(10), 2124–2132 (2012).
Schweitzer L, Geisler C, Pourhassan M et al. What is the best reference site for a single MRI slice to assess whole-body skeletal muscle and adipose tissue volumes in healthy adults?1. The American Journal of Clinical Nutrition 102(1), 58–65 (2015).
Santini F, Pansini M, Deligianni X, Caligiuri ME, Oei EHG. ESR Essentials: advanced MR safety in vulnerable patients—practice recommendations by the European Society for Magnetic Resonance in Medicine and Biology. Eur Radiol (2024).
Euser AM, Dekker FW, Cessie S le. A practical approach to Bland-Altman plots and variation coefficients for log transformed variables. Journal of Clinical Epidemiology 61(10), 978–982 (2008).
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1(8476), 307–310 (1986).
Bland JM, Altman DG. Applying the right statistics: analyses of measurement studies. Ultrasound in Obstetrics & Gynecology 22(1), 85–93 (2003).
Raunig DL, McShane LM, Pennello G et al. Quantitative imaging biomarkers: a review of statistical methods for technical performance assessment. Stat Methods Med Res 24(1), 27–67 (2015).
Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. Journal of Chiropractic Medicine 15(2), 155–163 (2016).
Lu M-J, Zhong W-H, Liu Y-X, Miao H-Z, Li Y-C, Ji M-H. Sample Size for Assessing Agreement between Two Methods of Measurement by Bland-Altman Method. Int J Biostat 12(2), /j/ijb.2016.12.issue-2/ijb-2015-0039/ijb-2015-0039.xml (2016).
Schaudinn A, Linder N, Garnov N et al. Predictive accuracy of single- and multi-slice MRI for the estimation of total visceral adipose tissue in overweight to severely obese patients. NMR in Biomedicine 28(5), 583–590 (2015).
Borga M, Ahlgren A, Romu T, Widholm P, Dahlqvist Leinhard O, West J. Reproducibility and repeatability of MRI-based body composition analysis. Magnetic Resonance in Medicine 84(6), 3146–3156 (2020).
Newman D, Kelly-Morland C, Leinhard OD et al. Test–retest reliability of rapid whole body and compartmental fat volume quantification on a widebore 3T MR system in normal-weight, overweight, and obese subjects. Journal of Magnetic Resonance Imaging 44(6), 1464–1473 (2016).
Middleton MS, Haufe W, Hooker J et al. Quantifying Abdominal Adipose Tissue and Thigh Muscle Volume and Hepatic Proton Density Fat Fraction: Repeatability and Accuracy of an MR Imaging-based, Semiautomated Analysis Method. Radiology 283(2), 438–449 (2017).
Ropponen A, Videman T, Battié MC. The reliability of paraspinal muscles composition measurements using routine spine MRI and their association with back function. Manual Therapy 13(4), 349–356 (2008).
Hu Z-J, He J, Zhao F-D, Fang X-Q, Zhou L-N, Fan S-W. An assessment of the intra- and inter-reliability of the lumbar paraspinal muscle parameters using CT scan and magnetic resonance imaging. Spine (Phila Pa 1976) 36(13), E868-874 (2011).
Gerard EL, Snow RC, Kennedy DN et al. Overall body fat and regional fat distribution in young women: quantification with MR imaging. AJR Am J Roentgenol 157(1), 99–104 (1991).
Silver HJ, Niswender KD, Kullberg J et al. Comparison of gross body fat-water magnetic resonance imaging at 3 Tesla to dual-energy X-ray absorptiometry in obese women. Obesity (Silver Spring) 21(4), 765–774 (2013).
Kullberg J, Johansson L, Ahlström H et al. Automated assessment of whole-body adipose tissue depots from continuously moving bed MRI: a feasibility study. J Magn Reson Imaging 30(1), 185–193 (2009).
Bonekamp S, Ghosh P, Crawford S et al. Quantitative comparison and evaluation of software packages for assessment of abdominal adipose tissue distribution by magnetic resonance imaging. Int J Obes 32(1), 100–111 (2008).
Kiefer LS, Fabian J, Lorbeer R et al. Inter- and intra-observer variability of an anatomical landmark-based, manual segmentation method by MRI for the assessment of skeletal muscle fat content and area in subjects from the general population. Br J Radiol 91(1089), 20180019 (2018).

Competing interest reported. MN, LN, CEH, TD, LFC, RS, HTB, and MDR are employees at Perspectum Ltd, the company that developed VitruvianScan. HTB and MDR are also shareholders at Perspectum Ltd. MP is a consultant for and a shareholder at Perspectum.

SupplementaryInformationRepeatabilityandReproducibilityofSingleSliceL3MRIinBodyCompositionAssessment.docx

Single-slice MRI for body composition assessment: repeatability, reproducibility, and observer variability

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods