Computational and Mathematical Methods in Medicine

Research Article

Differentiating Small (≤1 cm) Focal Liver Lesions as Metastases or Cysts by means of Computed Tomography: A Case-Study to Illustrate a Fuzzy Logic-Based Method to Assess the Impact of Diagnostic Confidence on Radiological Diagnosis

Table 6

Fuzzy performance of a simulated reader R3 showing low crisp accuracy for malignancy (metastases versus cysts at multidetector row Computed Tomography) and high diagnostic confidence (DC) in correct diagnoses. Simulations 1 and 2 assume that DC in incorrect diagnoses (false negatives and false positives) are expressed with high or low DC levels, respectively (see the text for further detail). In brackets we reported 95% C.I.s.


	Crisp	Simulation 1		Simulation 2
	Crisp	Fuzzy	(%)	Fuzzy	(%)

Sensitivity (%)	66.7	63.3		70.0
	(51.8–79.0)	(48.5–76.1)	+5.26	(55.2–81.7)	−4.76
	(20/30)	(19/30)		(21/30)

Specificity (%)	50.0	50.0		60.0
	(35.7–64.3)	(35.7–64.3)	0	(54.2–73.3)	−16.7
	(10/20)	(10/20)		(12/20)

PPV (%)	66.7	65.5		72.4
	(51.8–79.0)	(50.7–78.0)	+1.75	(57.7–83.7)	−7.94
	(20/30)	(19/29)		(21/29)

NPV (%)	50.0	47.6		57.1
	(35.7–64.3)	(33.5–62.1)	+5.00	(42.4–70.8)	−12.5
	(10/20)	(10/21)		(12/21)

Accuracy (%)	60.0	58.0		66.0
	(54.2–73.3)	(43.3–71.5)	+3.45	(51.1–78.4)	−9.09
	(30/50)	(29/50)		(33/50)