Related Experiment Video
Updated: Aug 10, 2025
06:36
Lower Limb Biomechanical Analysis of Healthy Participants
Published on: April 15, 2020
9.2K
Age-Related Reliability of B-Mode Analysis for Tailored Exosuit Assistance
Letizia Gionfrida1, Richard W Nuckols2, Conor J Walsh1
1Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, Science and Engineering Complex, 150 Western Ave, Boston, MA 02134, USA.
Sensors (Basel, Switzerland)
|February 11, 2023
Summary
This study developed a ResNet + 2x-LSTM model to track muscle fascicle lengths from ultrasound images in young and older adults. The model shows potential for personalized wearable robotic assistance, even with age-related physiological changes.
Area of Science:
- Biomedical Engineering
- Robotics
- Gerontology
Background:
- Personalized assistance in wearable robotics is crucial for maximizing user benefit.
- Ultrasound-based muscle fascicle tracking has been used for young individuals, but strategies for older adults with age-altered physiology are needed.
Purpose of the Study:
- To introduce and validate a ResNet + 2x-LSTM model for extracting muscle fascicle lengths from B-mode ultrasound images in both young and older adults.
- To assess the model's performance across different age groups and determine its suitability for developing individualized assistive strategies.
Main Methods:
- A ResNet + 2x-LSTM model was developed for fascicle length extraction.
- The model was trained on semimanually labeled B-mode ultrasound data from young (40,696 frames) and older adults (34,262 frames) of the medial gastrocnemius.
- Performance was evaluated using metrics like R², RMSE, MAPE, and aaDF.
Main Results:
- When trained on young adults, the model achieved R² = 0.85 for young and R² = 0.53 for older adults.
- When trained across all ages, the model achieved R² = 0.79.
- Despite age-related muscle loss, the absolute percentage error (3-5%) suggests acceptable accuracy for generating assistive force profiles.
Conclusions:
- The ResNet + 2x-LSTM model effectively extracts muscle fascicle lengths from ultrasound data in both young and older adults.
- The model's accuracy, even with age-related differences, supports its application in creating personalized assistive strategies for wearable robotics in diverse age groups.
