Title :
A Three-Dimensional Method for Detection of Pulmonary Nodule
Author :
Yang, Liu ; Liu Yang ; Li Wei ; Zhao Dazhe
Author_Institution :
Key Lab. of Med. Image Comput. of Minist. of Educ., Northeast Univ., Sheyang, China
Abstract :
A pulmonary nodule is relatively round lesion, or area of abnormal tissue located within the lung that can be seen in thoracic CT scans. Because noise and same like disturbance of blood vessels and tracheas, detection of the lung nodule is difficult. A three-dimensional pulmonary nodule detection method for thoracic CT scans is proposed in this paper. First, bounding box method and three-dimensional sphere-enhancement filter for nodule candidate selection are applied to enhance volume of interest (VOI). Then, 3D features of the VOI are extracted to train L the neural network classifier to reduce false positive rate. With this method, we can effectively decrease the noises which are nonsphere and achieve a low false positive rate.
Keywords :
blood vessels; computerised tomography; feature extraction; filtering theory; image classification; lung; medical image processing; neural nets; abnormal tissue; blood vessel; bounding box method; false positive rate reduction; feature extraction; lung lesion; neural network classifier; pulmonary nodule detection; thoracic CT scan; three-dimensional sphere-enhancement filter; trachea; Biomedical imaging; Blood vessels; Computed tomography; Filtering theory; Filters; Laboratories; Lungs; Machine learning; Neural networks; Shape;
Conference_Titel :
Biomedical Engineering and Informatics, 2009. BMEI '09. 2nd International Conference on
Conference_Location :
Tianjin
Print_ISBN :
978-1-4244-4132-7
Electronic_ISBN :
978-1-4244-4134-1
DOI :
10.1109/BMEI.2009.5304870
