Pancreatic cystic lesions: discrimination accuracy based on clinical data and high resolution CT features

J Comput Assist Tomogr. 2007 Nov-Dec;31(6):860-7. doi: 10.1097/RCT.0b013e318039b277.


Objective: To determine the accuracy of high-resolution computed tomography in discriminating among pancreatic cystic lesions with and without clinical history.

Methods: Multiphasic high-resolution computed tomographic scans of 100 proven cystic pancreatic lesions were retrospectively reviewed by 2 blinded imagers. Before and after a history of pancreatitis was revealed, readers classified lesions as either nontumor-related or tumor-related. Reviewers also classified lesions as benign or malignant and finally categorized each lesion as serous, mucinous, or other lesion. Diagnostic performance parameters for classification were calculated for each reader with and without history.

Results: Sixty-two lesions were tumor-related and 38 were not. Accuracy for classifying tumor-related from unrelated cysts for both reviewers increased (83%-92%, 85%-91%) when unblinded to pancreatitis history. Specificity also significantly increased (66%-88% and 84%) for both reviewers.

Conclusions: Pancreatic-phase computed tomography enabled more than 80% accuracy in discriminating between tumor-related and unrelated lesions and increased with information of pancreatitis history.

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Aged, 80 and over
  • Biopsy
  • Cystadenocarcinoma / diagnostic imaging
  • Cystadenoma / diagnostic imaging
  • Diagnosis, Differential
  • Female
  • Follow-Up Studies
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Male
  • Middle Aged
  • Pancreatic Cyst / diagnostic imaging*
  • Pancreatic Cyst / pathology
  • Pancreatic Neoplasms / diagnostic imaging*
  • Pancreatic Neoplasms / pathology
  • Pancreatic Pseudocyst / diagnostic imaging
  • Pancreatitis / complications
  • Precancerous Conditions / diagnostic imaging
  • Retrospective Studies
  • Sensitivity and Specificity
  • Single-Blind Method
  • Tomography, X-Ray Computed / methods*