This project will develop a computer assisted consultation system for clinical radiologists in the detection of lung nodules (cancer) on chest radiographs. An automated screenmg of digital chest radiographs can be further developed when this system is fully tested in various clinical settings. The success of this project will inspire revolutionary improvements in other cancer diagnostic procedures through further research and development. We have demonstrated that the newly developed "vision type" neural network and training methods for the detection of lung nodules are also applicable to the detection of microcalcifications on mammograms by presenting microcalcification patterns in the training. The pre scan methods and vision type neural networks simulate radiologists' routine practices m reading chest radiographs Radiologists' viewing patterns and decision making processes will be modeled and converted to computer readable form. Preliminary studies have shown the promise of this approach. The Phase I study will address issues related to the differentiation of end on vessels from true nodules. The plan of the Phase II study is to (i) analyze the learninz patterns of the vision type neural network, (ii) consolidate the research outcome of the neural network learning, and (iii) implement a prototype consultation system for clinical use.Commercial ApplicationsBecause of the low detection rate of lung nodules in clinical radiology, the commercial applications of this SBIR research are very promising. We expect that this system will be adapted to the analysis of chest radiography which is a dominant radiological procedure in the majority of clinics and hospitals.National Cancer Institute (NCI)
