Table 1 Segmentation Performance Metrics of the CNN on the Testing Dataset (n = 14).

From: Deep Learning Convolutional Neural Networks for the Automatic Quantification of Muscle Fat Infiltration Following Whiplash Injury

Performance Metric

Left

Right

Sørensen–Dice Index

0.862 ± 0.017

0.871 ± 0.016

Jaccard Index

0.758 ± 0.026

0.772 ± 0.026

Conformity Coefficient

0.678 ± 0.046

0.703 ± 0.044

True Positive Rate

0.904 ± 0.021

0.908 ± 0.017

True Negative Rate

0.999 ± < 0.001

0.999 ± < 0.001

Positive Predictive Value

0.829 ± 0.031

0.843 ± 0.032

Volume Ratio

1.100 ± 0.058

1.087 ± 0.058

  1. Metrics shown = average ± 1 standard deviation.
Back to article page