Automatic Segmentation and Measurement on Knee Computerized Tomography Images for Patellar Dislocation Diagnosis
Table 1
Mean (MEAN) and standard deviation (SDEV) statistics of overlap rate (), false-positive rate (), and Dice similarity coefficient () and mean statistics of separation rate () for different image segmentation methods (C-V, BCFCM, RSF, LSEBFE, and LIF) with and without our framework on all the test CT images.
With our framework?
Methods
Performance metric
Yes
C-V
95.09 ± 3.24
2.04 ± 2.70
0.965 ± 0.022
100.00
BCFCM
99.12 ± 2.48
6.32 ± 5.63
0.962 ± 0.035
100.00
RSF
93.69 ± 4.34
1.79 ± 2.33
0.958 ± 0.025
100.00
LSEBFE
94.60 ± 3.96
1.85 ± 2.09
0.963 ± 0.023
100.00
LIF
94.12 ± 3.83
9.43 ± 10.74
0.919 ± 0.063
100.00
No
C-V
95.04 ± 3.90
3.24 ± 3.43
0.958 ± 0.026
25.33
BCFCM
81.88 ± 10.01
6.47 ± 12.03
0.862 ± 0.081
54.67
RSF
69.66 ± 9.43
3.05 ± 3.81
0.807 ± 0.069
68.67
LSEBFE
82.64 ± 7.55
2.76 ± 4.19
0.892 ± 0.055
55.33
LIF
73.79 ± 7.13
11.18 ± 9.53
0.802 ± 0.062
68.67
Numeric entries for ,, and in the table are in the form of MEAN ± SDEV.
