Pedagogic Tool for Learn-Teaching on Digital Mammography Using Dedicated Software

摘要

The purpose of this work is to evaluate the effectiveness of a training methodology developed in the Medical Image Qualification Laboratory - QuallM to assist in the complex learning process of reading mammography images. A dedicated software was used by residents and specialists for image manipulation (contrast, brightness, zoom, inversion) and also to analyze the readings performance. Sixty images of statistical phantom (Alvim?) were acquired with mammo-graphic equipment using different techniques and presented in monitors and a specific view box (screen film). The readings' performance was evaluated by statistical parameters: probability of detectabiliry (Pdet), kappa (K), true positive results (TP), false positive results (FP) and the area under the ROC curve, considering different objects (fibers and microcalcifica-tions) and their different dimensions. The residents' detection performance was superior in monitors (softcopy) for both objects (fibers and microcalcifications) suggesting that digital tools were useful since they promoted higher contrast resolution. The ROC analysis show a higher performance in the digital system (Az=0.808) than in the screen film system (Az= 0.707). Also the experts got better results in the digital system (Az=0.841) than with another reading system (Az=0.803). The methodology allowed to quantify and compare the readings performance among different groups based on Signal Detection Theory.