Analysis of a three-dimensional lung nodule detection method for thoracic CT scans

摘要: We are developing an automated method to analyze the three- dimensional nature of structures within CT scans and identify those that represent lung nodules. The set segmented regions from all sections a scan forms volume which multiple gray-level thresholds applied. Contiguous three-dimensional identified each thresholded volume, satisfy criterion constitute initial nodule candidates. A feature vector is then computed for candidate. rule-based scheme applied candidate reduce number candidates correspond normal anatomy. Feature vectors remaining merged through classifier further distinguish between nodules structures. This demonstrates promising performance in its ability detect images. Such technique may assist radiologists evaluate, example, images low-dose, screening thoracic examinations.© (2000) COPYRIGHT SPIE--The International Society Optical Engineering. Downloading abstract permitted personal use only.