研究文章
糖尿病性视网膜病变检测使用本地极值量子化Haralick特性与长期短期记忆网络
|
| 图像 |
特异性(%) |
灵敏度(%) |
SSI (%) |
PPV (%) |
NPV (%) |
精度(%) |
|
| Image01 |
94.34 |
70.54 |
99.38 |
55.31 |
96.99 |
92.19 |
| Image02 |
82.03 |
91.84 |
98.55 |
27.67 |
99.26 |
82.71 |
| Image03 |
92.88 |
86.88 |
99.47 |
51.54 |
98.78 |
92.40 |
| Image04 |
92.79 |
78.70 |
99.37 |
46.83 |
98.18 |
91.74 |
| Image05 |
94.13 |
85.30 |
99.55 |
55.36 |
98.68 |
93.43 |
|
|
