PRACTICAL ONCOLOGY JOURNAL ›› 2011, Vol. 25 ›› Issue (2): 126-128.doi: 10.3969/j.issn.1002-3070.2011.02.006

Previous Articles     Next Articles

The neural network model used for identified benign and malignant of the solitary pulmonary nodules

SUN Li1, ZHANG Xiushi1, ZHOU Yang1, LIU Lu2   

  1. 1.The Third Affiliated Hospital of Harbin Medical University,Harbin 150081;
    2.Harbin University of Science and Technology
  • Received:2010-11-03 Online:2011-02-25 Published:2012-02-21

PDF (PC)

50

Abstract: Objective To evaluate the significance of diagnosis of isolation pulmonary nodules(SPN)using computer-aided diagnosis(CAD).Method 193 cases with SPN(including 144 cases of malignant nodules and 49 cases of benign nodules)proved pathologically by CT-guided biopsy in this paper,and analyzed the characteristics.According to the pathology results,Compare the accuracy of the neural network and artificial diagnosis of two methods in SPN's diagnostic.Result The diagnosis accurancy of neural network model was 71.5%,and the accurancy of artificial diagnosis was 80.3%,Sensitivity was 84.0% and 91.0%,respectively,specificity was 34.7% and 49.0% respectively.Conclusion Computer-aided diagnosis has better diagnostic accuracy rate,and the sensitivity of artificial neural network was closed to physicians diagnosis.It has certain significance in SPN diagnosis,can be a good diagnostic value,but with less accuracy.So it cannot completely replace the artificial diagnosis,but can be used as auxiliary diagnostic.

Key words: Isolation pulmonary nodules(SPN), Computer-aided diagnosis(CAD), Artificial neural network

CLC Number: