The　high　complexity　of　intravascular　ultrasound　(IVUS)　image　determines　that　the　appropriate　processing　method　for　IVUS　image　is　pivotal.　In　this　paper　a　new　way　is　used　to　reduce　the　complexity　in　IVUS　image　preprocessing.　At　first　the　IVUS　original　image　is　preprocessed　with　the　polar　coordinates　transform　and　the　threshold　segmentation　based　on　entropy　for　gray-level　picture,　or　binary　image　processing.　The　noise　dots　are　removed　before　GVF　transform.　Then　the　preprocessed　image　is　transformed　by　GVF.　Finally,　the　desirable　lumen　and　media-adventitial　borders　of　IVUS　image　are　obtained.　The　experimental　results　confirm　the　validity　of　the　new　method.