Introduction
Materials and methods
Study cohort
[99mTc]TcDTPA SPECT/CT scan acquisition
Annotations for EOM contours and GO activity
Data preprocessing
Deep learning model architecture
Evaluation
Results
Segmentation performance
SR | LR | MR | IR | Average | |
---|---|---|---|---|---|
IOU | 0.79 ± 0.03 | 0.81 ± 0.06 | 0.85 ± 0.08 | 0.82 ± 0.08 | 0.82 ± 0.05 |
SP (%) | 99.93 ± 0.04 | 99.97 ± 0.02 | 99.96 ± 0.03 | 99.96 ± 0.03 | 99.96 ± 0.03 |
SN (%) | 89.13 ± 3.92 | 90.81 ± 4.71 | 92.35 ± 7.52 | 90.62 ± 7.93 | 91.01 ± 5.32 |
Classification performance
Accuracy(%) | Precision (%) | Sensitivity (%) | Specificity (%) | F1 score | AUC | |
---|---|---|---|---|---|---|
CT | 54.92 | 55.88 | 57.14 | 53.68 | 0.47 | 0.54 |
CT + SPECT | 60.88 | 58.56 | 51.58 | 75.51 | 0.63 | 0.67 |
CT + EOM mask | 75.13 | 77.65 | 74.74 | 87.76 | 0.81 | 0.85 |
CT + SPECT + EOM masks | 86.10 | 85.17 | 87.37 | 85.71 | 0.86 | 0.90 |
Accuracy (%) | Precision (%) | Sensitivity (%) | Specificity (%) | F1 score | AUC | |
---|---|---|---|---|---|---|
Training set | 84.31 ± 0.25 | 85.67 ± 1.06 | 86.64 ± 0.42 | 82.32 ± 1.28 | 0.85 ± 0.01 | 0.90 ± 0.01 |
Validation set | 82.06 ± 0.32 | 85.01 ± 2.35 | 83.63 ± 1.54 | 82.53 ± 1.26 | 0.83 ± 0.02 | 0.89 ± 0.01 |
Test set | 84.25 ± 1.25 | 83.35 ± 1.53 | 84.63 ± 0.84 | 83.87 ± 1.97 | 0.83 ± 0.01 | 0.89 ± 0.02 |