Visual perceptual learning of mammographic lesions

Human subjects (n = 72, college-age) without any medical experience were trained to detect different types of mammographic lesions indicative of breast cancer (calcification lesion, distortion lesion) among normal mammograms. The content of feedback about response accuracy after each trial of training was varied between different subject groups. Specifically, subjects either received detailed feedback, or partial feedback, or no feedback about response accuracy after each trial of training. There was a total of n = 12 subjects for each combination of trained lesion (calcification, distortion) and content of response feedback (detailed, partial, none). Subjects completed a pretest, three training sessions and a posttest. Furthermore, a subset of trained subjects was available for a retest conducted six months after the posttest. Each session was conducted on a separate day. The results show that learning from pretest to posttest occurred for both types of lesions with detailed feedback during training. Learning also occurred for calcification lesions with partial feedback. Learning for both types of lesions was retained during the retest with detailed feedback during training.