Reliability and Validity of Muscle Size and Quality Analysis Techniques

Area of Science:

• Musculoskeletal ultrasound imaging
• Biomechanical analysis
• Image processing algorithms

Background:

• Accurate measurement of muscle cross-sectional area (ACSA) and echo intensity (EI) is crucial for assessing muscle health and function.
• Manual analysis of ultrasound images can be time-consuming and subject to inter-observer variability.
• Automated image analysis tools aim to improve efficiency and consistency in these measurements.

Purpose of the Study:

• To evaluate the reliability and validity of an automatic image analysis program, Deep Anatomical Cross-Sectional Area (DeepACSA), for measuring vastus lateralis (VL) muscle cross-sectional area (ACSA) and echo intensity (EI).
• To compare the performance of the DeepACSA program against manual analysis methods.

Main Methods:

• Twenty-two participants underwent ultrasound imaging of the VL muscle at 10 and 12 MHz across two separate trials.
• Muscle images were analyzed both manually and using the DeepACSA software.
• Reliability (intraclass correlation coefficient [ICC], standard error of measure [SEM%]) and validity (constant error [CE], total error [TE], standard error of the estimate [SEE]) statistics were calculated.

Main Results:

• Automatic ACSA analysis demonstrated good reliability (ICC: 0.83-0.90 at 10 MHz, 0.87-0.88 at 12 MHz), while manual analysis showed moderate to excellent reliability (ICC: 0.82-0.99 at 10 MHz, 0.73-0.99 at 12 MHz).
• Automatic EI analysis also showed good reliability, comparable to manual methods.
• The DeepACSA program exhibited lower error for ACSA at 12 MHz (TE=3.44 cm², SEE=3.11 cm²) compared to 10 MHz (TE=4.94 cm², SEE=3.65 cm²).
• Error for EI analysis was slightly lower at 12 MHz (TE=2.61 a.u., SEE=2.34 a.u.) compared to 10 MHz (TE=2.70 a.u., SEE=2.58 a.u.).

Conclusions:

• The DeepACSA program shows potential for reliable and low-error measurement of VL muscle EI, comparable to manual analysis.
• While DeepACSA demonstrated good reliability for ACSA, manual analysis may still be preferred for higher precision.
• The automatic program exhibits low overall error for both ACSA and EI measurements at 10 and 12 MHz ultrasound frequencies.