Research Article

New Morphological Features for Grading Pancreatic Ductal Adenocarcinomas

Table 8

Evaluation results for distinguishing Normal and PDAC to each feature set.

ObjectFeature setTNFPFNTPSN (%)SP (%)PPV (%)NPV (%)ACC (%)

NENCNF
(1.58)(2.11)(1.58)(2.11)(4.94)(6.60)(5.17)(4.22)(3.95)

ā€ƒCLF
LumenPLF29.302.702.7029.3091.5791.5791.9991.8891.56
(1.77)(2.00)(1.77)(2.00)(5.52)(6.26)(5.45)(4.70)(3.14)
ā€ƒALF

ā€ƒCEF
ENPEF27.503.504.5028.5085.9489.0789.5286.6787.50
(1.51)(2.95)(1.51)(2.95)(4.72)(9.23)(6.80)(3.25)(3.21)
ā€ƒAEF

ā€ƒCDF
DuctPDF29.801.402.2030.6093.1395.6395.7893.5094.38
(1.40)(1.51)(1.40)(1.51)(4.37)(4.71)(4.34)(3.80)(2.35)
ā€ƒADF

ā€ƒCTF
TissuePTF29.802.502.2029.5093.1392.1992.7493.2692.66
(1.32)(2.22)(1.32)(2.22)(4.11)(6.95)(6.15)(3.76)(3.21)
ā€ƒATF

TP (true positive): the number of PDACs that are correctly classified as PDACs.
FP (false positive): the number of Normals that are incorrectly classified as PDACs.
FN (false negative): the number of PDACs that are incorrectly classified as Normals.
TN (true negative): the number of Normals that are correctly classified as Normals.