The reliability and validity of triceps surae muscle volume assessment using freehand three-dimensional ultrasound in typically developing infants

Abstract

This study assessed the intra-acquirer, intra- and inter-processor reliability, and validity of the in vivo assessment of the medial gastrocnemius (MG), lateral gastrocnemius (LG) and soleus (SOL) muscle volumes using freehand 3D ultrasound (3DUS) in typically developing infants. Reliability assessments of freehand 3DUS were undertaken in infants across three ages groups: three, six and twelve months of age, with validity testing completed against magnetic resonance imaging (MRI) in infants at 3 months of age. Freehand 3DUS scanning was carried out by a single acquirer, with two independent processors manually segmenting images to render volumes. MRI images were segmented independently by a separate processor, with the volumes compared to those obtained via freehand 3DUS. Reliability was assessed using intraclass correlation (ICC), coefficient of variance (CV) and minimal detectable change (MDC) across each assessment time point. Validity was assessed using the limits of agreement. ICCs for intra-acquirer reliability of the acquisition process for freehand 3DUS ranged from 0.91 to 0.99 across all muscles. ICCs for intra-processor and inter-processor reliability for the segmentation process of freehand 3DUS ranged from 0.80 to 0.98 across all muscles. Acceptable levels of agreement between muscle volume obtained by freehand 3DUS and MRI were found for all muscles; however, freehand 3DUS overestimated muscle volume of MG and LG and underestimate the SOL compared with MRI, with average absolute differences of MG = 0.3 ml, LG = 0.3 ml and Sol = 1.2 ml. Freehand 3DUS is a reliable method for measuring in vivo triceps surae muscle volume in typically developing infants. We conclude that freehand 3DUS is a useful tool to assess changes in muscle volume in response to growth and interventions in infants.

Keywords: child; muscle volume; reliability; statistical shape modelling; ultrasound; validity.

FIGURE 1

The experimental design. From a larger group of 23 typically developing infants, data from 10 infants were included for each analysis

FIGURE 2

One participant position for 3DUS scanning

FIGURE 3

Muscle volume assessment of all three muscles with the freehand 3DUS method in a three‐month‐old participant. A B‐mode ultrasound image of the muscle before segmentation (left). A single B‐mode ultrasound image after manual segmentation (middle). Reconstructed muscle (right)

FIGURE 4

MRI slice of a cross‐section of the left lower limb of a three‐month‐old participant. Before segmentation (left) After segmentation (right) demonstrating each of the three muscles: MG (blue), LG (yellow), and SOL (pink)

FIGURE 5

This study established the intra‐acquirer and inter and intra‐processor reliability of freehand three‐dimensional (3D) ultrasound to measure in vivo muscle volume in infants 3–12 months of age, as well as the validity (compared with MRI) in 3‐month‐old infants. The spatial error map developed from our statistical shape analysis showed the sources of segmentation error for both the Lateral and Medial Gastrocnemius, and Soleus muscle between two acquirers

FIGURE 6

Bland–Altman Graphs, absolute difference (3DUS—MRI) vs. average of values measured by MRI and freehand 3DUS (left) and scatter plots (MRI vs. 3DUS) of the MG, LG, and SOL (right). The horizontal lines on the Bland–Altman plots represent the mean difference (solid) and the upper and lower 95% limits of agreement (dashed). The dashed diagonal line in the scatterplots represents the line of perfect agreement