ERCP diagnosis of tumors in the region of the head of the pancreas. Analysis of criteria and computer-aided diagnosis

Dig Dis Sci. 1988 May;33(5):577-86. doi: 10.1007/BF01798360.

Abstract

During the period 1974-1983, 320 patients with pancreas carcinoma, papilla of Vater carcinoma, bile duct bifurcation carcinoma, or duodenal carcinoma were examined by ERCP. Using 30 ERCP criteria, a radiological diagnosis was made. A valid pathological diagnosis was available in 200 patients (62.5%). In 183 of the 200 patients (91.5%), the ERCP diagnosis and the pathological diagnosis were identical. We then performed an analysis using 52 ERCP criteria. In 192 of the 200 patients (96.0%), the ERCP diagnosis based on this reanalysis and the pathological diagnosis were identical. By discriminant analysis, 13 ERCP criteria with a maximal discriminatory value were selected in patients in whom all diagnostic structures (bile ducts, pancreatic duct, and duodenum) were visible. Using these 13 criteria selected by discriminant analysis, the diagnostic score was 98.9%. A computer program based on these 13 ERCP criteria was designed for use in practice. The diagnostic accuracy of this computer program was 98.4%. Finally we tested this computer program on 171 new patients who were seen in the period 1983-1986. In 143 of the 171 patients, a valid pathological diagnosis was available (83.6%). Comparing the ERCP diagnosis in all patients (even if not all structures were visible) with a valid pathological diagnosis, the prospective score of the computer program was 91.6%. Using this program it was possible to evaluate examinations in which not all structures were visible. When the ERCP diagnosis was uncertain, the doubt could be quantified.

Publication types

  • Comparative Study

MeSH terms

  • Ampulla of Vater / diagnostic imaging
  • Bile Duct Neoplasms / diagnostic imaging
  • Cholangiopancreatography, Endoscopic Retrograde*
  • Common Bile Duct Neoplasms / diagnostic imaging
  • Diagnosis, Differential
  • Duodenal Neoplasms / diagnostic imaging
  • Humans
  • Image Interpretation, Computer-Assisted*
  • Pancreatic Neoplasms / diagnostic imaging*
  • Radiographic Image Interpretation, Computer-Assisted*
  • Software