A New Real-Time Method for Lung Mass Lesion Detection in CT Images

Detection of lung mass lesion is widely required in the clinic for diagnosis of different pulmonary defects like lung cancer and pulmonary embolism. Due to severe symptoms of such diseases, real-time detection of the lung mass lesion is significant. In this paper, this problem is addressed by a new image segmentation algorithm. Primarily, a Gabor-based filtering algorithm is employed to remove the intensity nonuniformity of the CT image. Then, the holes and discontinuities of the mass lesion are eliminated through morphological operations. Finally, the lesion objects are extracted by using the thresholding and connected-component-analysis. The experimental results demonstrated significantly short CPU time of the proposed algorithm, as 73 milliseconds for each slice on a typical laptop. Also, the solution quality of our method was considerably high, as 91.6%, 85.4%, and 93.3% in terms of accuracy, sensitivity, and specificity, respectively. We further showed that the proposed method provided better solutions compared to six other counterpart algorithms.