• DocumentCode
    3227175
  • Title

    Computer-Aided UltraSound (CAUS) to determine liver fat content in animal model and in patients

  • Author

    Weijers, G. ; Wanten, G. ; Thijssen, J.M. ; van der Graaf, M. ; de Korte, C.L.

  • Author_Institution
    Dept. of Pediatrics, Radboud Univ. Nijmegen Med. Centre, Nijmegen, Netherlands
  • fYear
    2011
  • fDate
    18-21 Oct. 2011
  • Firstpage
    41
  • Lastpage
    44
  • Abstract
    The authors developed and tested a quantitative ultrasound technique (CAUS) for the staging of steatosis using Ultrasound Tissue Characteristics Parameters (UTCPs) from B-mode images [1]. CAUS method pre-processing steps in short: transformation of the gray level to logarithmic echo level by linearization of the post-pocessing Look Up Table (LUT); Back-Scan conversion of sector image format, i.e., transformation of carthesian DICOM into polar data, in order to linearize and rotate speckles horizontally; depth dependent gray level corrections, i.e., correction for the combined skin, fat and muscle layer (fat layer), automatic gain correction (AGC) by using gray level depth profile in phantom; attenuation correction, and finally automatic image segmentation for the removal of medium and large sized hepatic blood vessels and bile ducts. The following UTCPs were then estimated; Mean echo level (Mu), Standard Deviation of mean echo level (SD), Signal to Noise Ratio (SNR), Residual Attenuation coefficient (ResAtt), Axial (Ax) and Lateral (Lat) speckle size. This method was recently validated with an extensive animal study in 151 post-partem cows [2, 3]. Since also biopsies were taken, correlation with triacylglycerol (TAG) content could be investigated. High UTCP correlations were found for Mean echo level (Mu, r=0.7), Residual Attenuation coefficient (ResAtt, r=0.81), Signal to Noise Radio (SNR, r=0.74) with triacylglycerol (TAG) content. ROC analysis revealed high values for Sensitivity (87%) and Specificity (87%) for the classification into different risk groups. In the present study, data were obtained from 13 human patients (3 males) on Home Parenteral Nutrition (HPN, administrated at least 6 times per week) with a mean age of 42 yrs. Five independent intercostal US liver images per patient were acquired using a Sonos7500 (Philips Ultrasound, Andover, MA, USA) with a phased array transducer (S8, bandwidth 3-8 MHz). On all US patient data the CAUS method was performe- and the results were validated using a quantification of liver fat concentration by proton Magnetic Resonance Spectroscopy (MRS, 3T Trio MR system, Siemens, Erlangen, Germany) at the same day. Again high correlation values with fat level were obtained for several UTCPs: Mu r=0.78, ResAtt r=0.88, SNR r=0.91, Lat r=-0.62. In conclusion, CAUS parameters may has potential in staging, and possibly even screening, of hepatic steatosis in human liver, thus preventing taking liver biopsies.
  • Keywords
    CAD; biomedical MRI; biomedical ultrasonics; blood; blood vessels; image segmentation; liver; medical image processing; muscle; phantoms; sensitivity; skin; ultrasonic imaging; ultrasonic transducer arrays; B-mode images; animal model; automatic gain correction; automatic image segmentation; axial speckle size; back-scan conversion; bile ducts; biopsy; carthesian DICOM-polar data transformation; computer-aided ultrasound; frequency 3 MHz to 8 MHz; gray level depth profile; hepatic blood vessels; hepatic steatosis; home parenteral nutrition; independent intercostal US liver images; lateral speckle size; liver fat content; logarithmic echo level; mean echo level standard deviation; muscle; phantom; phased array transducer; post-partem cows; post-pocessing look up table; proton magnetic resonance spectroscopy; quantitative ultrasound technique; residual attenuation coefficient; sector image format; sensitivity; signal-to-noise ratio; skin; triacylglycerol; ultrasound tissue characteristics parameters; Attenuation; Correlation; Image segmentation; Liver; Signal to noise ratio; Speckle; Ultrasonic imaging; Attenuation; Echo Level; Liver; NAFLD; Quantitative; Speckle Size; Steatosis; Ultrasound;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Ultrasonics Symposium (IUS), 2011 IEEE International
  • Conference_Location
    Orlando, FL
  • ISSN
    1948-5719
  • Print_ISBN
    978-1-4577-1253-1
  • Type

    conf

  • DOI
    10.1109/ULTSYM.2011.0011
  • Filename
    6293264