Table 6 Performance comparison of several existing computer aided detection systems which used JSRT Database
From: The detection of lung cancer using massive artificial neural network based on soft tissue technique
Author | Sensitivity | FPs/image | Methodology | Classifier | Database |
|---|---|---|---|---|---|
Wei et al. [48] | 80% (123/154) | 5.4 (1333/247) (less accuracy due to high FPs) | Forward stepwise selection | Fisher linear discriminant | All abnormal as well as normal image inside JSRT(247) |
Coppini et al. [49] | 60% (93/154) | 4.3 (662/154) | Neural network filter | Fisher linear discriminant | All nodule image in JSRT(154) |
Schilham et al. [13] | 51% (79/154) 67% (103/154) | 2 (308/154) 4 (616/154) | Image filtering Based on Regression | Fisher linear discriminant | All nodule image in JSRT(154) |
Hardie et al. [50] | 80% (112/140) 63% (88/140) | 5.0 (700/140) 2 (280/140) | Active shape model and new weighted multi- scale conver gence-index | Fisher linear discriminant | Nodule image in JSRT(140) |
Chen et al. [31] | 71% (100/140) | 2 (466/233) | Computer aided detection using neural filter | Support Vector Machine (SVM) | Nodule and Normal image in JSRT(233) |
Proposed MANN based soft tissue technique | 72.85% (102/140) | 1 (233/233) | MANN for rib suppression | Support Vector Machine (SVM) | Nodule and normal image in JSRT(233) |