Radiological Physics and Technology

, Volume 2, Issue 2, pp 175–182

Application of an artificial neural network to the computer-aided differentiation of focal liver disease in MR imaging

Authors

    • Department of Electronics and Information Engineering, School of Computer, Electronics and InformationGuangxi University
  • Masayuki Kanematsu
    • Department of RadiologyGifu University School of Medicine
  • Hiroshi Fujita
    • Division of Regeneration and Advanced Medical Sciences, Department of Intelligent Image Information, Graduate School of MedicineGifu University
  • Xiangrong Zhou
    • Division of Regeneration and Advanced Medical Sciences, Department of Intelligent Image Information, Graduate School of MedicineGifu University
  • Takeshi Hara
    • Division of Regeneration and Advanced Medical Sciences, Department of Intelligent Image Information, Graduate School of MedicineGifu University
  • Ryujiro Yokoyama
    • Department of RadiologyGifu University School of Medicine
  • Hiroaki Hoshi
    • Department of RadiologyGifu University School of Medicine
Article

DOI: 10.1007/s12194-009-0062-5

Cite this article as:
Zhang, X., Kanematsu, M., Fujita, H. et al. Radiol Phys Technol (2009) 2: 175. doi:10.1007/s12194-009-0062-5

Abstract

The differentiation of focal liver lesions in magnetic resonance (MR) imaging is primarily based on the intensity and homogeneity of lesions with different imaging sequences. However, these imaging findings are falsely interpreted in some patients because of the complexities involved. Our aim is to establish a computer-aided diagnosis system named LiverANN for classifying the pathologies of focal liver lesions into five categories using the artificial neural network (ANN) technique. On each MR image, a region of interest (ROI) in the focal liver lesion was delineated by a radiologist. The intensity and homogeneity within the ROI were calculated automatically, producing numerical data that were analyzed by feeding them into the LiverANN as inputs. Outputs were the following five pathologic categories of hepatic disease: hepatic cyst, hepatocellular carcinoma, dysplasia in cirrhosis, cavernous hemangioma, and metastasis. Of the 320 MR images obtained from 80 patients (four images per patient) with liver lesions, our LiverANN classified 50 cases of a training set into five types of liver lesions with a training accuracy of 100% and 30 test cases with a testing accuracy of 93%. The experiment demonstrated that our LiverANN, which functions as a computer-aided differentiation tool, can provide radiologists with a second opinion during the radiologic diagnostic procedure.

Keywords

MR imaging Focal liver disease Differentiation Artificial neural network Computer-aided diagnosis (CAD)

Copyright information

© Japanese Society of Radiological Technology and Japan Society of Medical Physics 2009