RESUMO
PURPOSE: The objective is to study mammography practice from an optimisation point of view by assessing the impact of simple and immediately implementable corrective actions on image quality. MATERIALS AND METHODS: This prospective multinational study included 54 mammography units in 17 countries. More than 21,000 mammography images were evaluated using a three-level image quality scoring system. Following initial assessment, appropriate corrective actions were implemented and image quality was re-assessed in 24 units. RESULTS: The fraction of images that were considered acceptable without any remark in the first phase (before the implementation of corrective actions) was 70% and 75% for cranio-caudal and medio-lateral oblique projections, respectively. The main causes for poor image quality before corrective actions were related to film processing, damaged or scratched image receptors, or film-screen combinations that are not spectrally matched, inappropriate radiographic techniques and lack of training. Average glandular dose to a standard breast was 1.5 mGy (mean and range 0.59-3.2 mGy). After optimisation the frequency of poor quality images decreased, but the relative contributions of the various causes remained similar. Image quality improvements following appropriate corrective actions were up to 50 percentage points in some facilities. CONCLUSIONS: Poor image quality is a major source of unnecessary radiation dose to the breast. An increased awareness of good quality mammograms is of particular importance for countries that are moving towards introduction of population-based screening programmes. The study demonstrated how simple and low-cost measures can be a valuable tool in improving of image quality in mammography.