EVALUATION OF AN EXPERT-SYSTEM ON LYMPH-NODE PATHOLOGY

Background: Pathfinder is an Expert System that assists pathologists i n making accurate diagnoses in the domain of lymph-node pathology. Pat hfinder provides a differential diagnosis based on the initial histolo gical feature(s) observed by the pathologist, and suggests to the user additional histological features for observation that are likely to n arrow the differential diagnosis. Purpose: To evaluate the diagnostic accuracy of pathologists with and without the Pathfinder. Methods: Thi rty H&E stained slides from 30 lymph node biopsy specimens on which a Consensus diagnosis was made by two experts were reviewed by 19 pathol ogists to evaluate Pathfinder. After a period of training, 10 patholog ists using Pathfinder (Interactive Computer Method) and 9 pathologists using the Routine Method (diagnosis without computer) determined a di fferential diagnosis for 15 slides (Test 1). Pathologists were then cr ossed over, trained, and evaluated the remaining 15 slides (Test 2). F or each test, the proportion of ''correct'' diagnoses was compared bet ween methods. In addition, the information integration attributes (mak ing logical diagnosis given a set of specific histologic features) of Pathfinder and pathologists were compared. Finally, feature identifica tion and quantification skills of pathologists were determined and cor related with the percent correct diagnosis. Results: The diagnostic ac curacy using Pathfinder was greater than that using the Routine Method (40% v 32%, P = .02). Diagnostic accuracy for the group of pathologis ts who made diagnosis using the Routine Method in Test 1 increased whe n they made diagnosis using Pathfinder in Test 2 (27% to 44%, P < .000 1). The proportion of correct diagnosis for the group of pathologists who used Pathfinder in Test 1 remained virtually unchanged when they g ave up Pathfinder in Test 2 (35% to 37%). The percentage of incompatib le feature identification (''atypical proliferation'' diagnosis) was s ignificantly lower after using Pathfinder (P < .0001). In addition, in formation integration attributes of Pathfinder were significantly supe rior than that of the pathologists (P < .0001). Conclusions: Pathfinde r is a valuable tool that assists pathologists in making accurate diag nosis because it has superior attributes than pathologists to integrat e information and to screen for observations incompatible with any spe cific disease. Copyright (C) 1997 by W.B. Saunders Company.