Skip to main content
. 2023 Feb 3;23(3):1670. doi: 10.3390/s23031670

Table 1.

Comparison of the performances of the two proposed ResNet + 2x-LSTM algorithms for young-age-trained and all-age-trained models. The table indicates the value of the coefficient of determination (R2), the root-mean-square error (RMSE) in millimeters (mm), the mean absolute percentage error with respect to the total length of the fascicles (in %), the average absolute frame difference (aaFD) of the individual subsegments in mm, and the average detection time (mean in seconds ± SD).

Performance
Metrics
Young-Age-Trained All-Age-Trained
Young Older Older
R2 0.85 0.53 0.79
RMSE (in mm) 2.36 ±1.51 4.7 ± 4.66 3.95 ± 2.51
MAPE (%) 3.69 5.19 4.5
aaFD (in mm) 0.48± 1.10 1.9 ± 1.39 0.67± 1.8
Average detection time (s) 0.59 ± 1.53 0.62 ± 0.32