Table 2 The comparison of different method in lung nodule segmentation.
From: Automatic detect lung node with deep learning in segmentation and imbalance data labeling
Evaluation | Without pre-processing | With pre-processing (mono input positive) | With pre-processing (mono input negative) |
|---|---|---|---|
Dice coefficient | 0.573 | 0.790 | 0.701 |
Confusion matrix | Accuracy:0.9986 | Accuracy:0.9997 | Accuracy:0.9997 |
Sensitivity:0.9033 | Sensitivity:0.9614 | Sensitivity:0.9724 | |
Specificity:0.9987 | Specificity:0.9998 | Specificity:0.9997 |
Evaluation | With pre-processing (hybrid positive output) | With pre-processing (hybrid negative output) | |
|---|---|---|---|
Dice coefficient | 0.744 | 0.639 | |
Confusion matrix | Accuracy:0.9996 | Accuracy:0.9994 | |
Sensitivity:0.9502 | Sensitivity:0.9650 | ||
Specificity:0.9996 | Specificity:0.9994 |
