Jr. Fielding et al., SPIRAL CT IN THE EVALUATION OF FLANK PAIN - OVERALL ACCURACY AND FEATURE ANALYSIS, Journal of computer assisted tomography, 21(4), 1997, pp. 635-638
Purpose: Our goal was to assess test reliability and identify those fe
atures that have the strongest positive and negative predictive values
in the diagnosis of renal colic using spiral CT. Method: Fifty non-co
ntrast-enhanced CT scans (5 mm slice thickness) obtained in patients p
resenting with flank pain were reviewed by three radiologists blinded
to the final diagnoses. The sensitivity, specificity, and positive and
negative predictive values for nine pertinent findings were determine
d as compared to clinical follow-up. Results: Twenty-nine cases had fi
ndings of ureteral obstruction. Findings with the strongest positive p
redictive values (>0.90) were ureteral stone, hydronephrosis, hydroure
ter, periureteral stranding, and ureterovesical junction edema. Findin
gs with the strongest negative predictive values (>0.89) were absence
of hydronephrosis and hydroureter. The areas under the receiver operat
ing curves for Readers 1, 2, and 3 were 0.970 +/- 0.030, 0.942 +/- 0.0
36, and 0.982 +/- 0.020. Conclusion: Absence of hydroureter and hydron
ephrosis on spiral CT images should prompt a search for a diagnosis ot
her than an obstructing ureteral stone.