[The simulation of histopathological diagnosis with the help of an automated image analysis system (VISIAC)]

Simulation of the diagnostic algorithm in histopathology at 2 different light microscopical magnifications (x4, x40) for separating 7 diagnostic groups in lung pathology (healthy lung, healthy bronchus, lymphocytic inflammation, epidermoid-, adeno-, small cell, and large cell carcinoma) is presented. 140 cases were tested as a teaching set, 105 cases as learning set. Gray value distribution discriminates between healthy parenchyma and the diseases in all cases tested; minimum spanning tree discriminates between inflammation and carcinoma in all cases, and in between the carcinoma with 70% to 80% accuracy. The measurements may be used as additional aid in difficult diagnostic cases.