Computer-aided Diagnosis for Detection of Lacunar Infarcts in MR Images

概要

論文の詳細を見る
The detection of asymptomatic lacunar infarcts in MRI images is an important task for radiologists to prevent the occurrence of serious cerebral infarctions. However, it is difficult for radiologists to identify the lacunar infarcts correctly in MRI images. Therefore, we developed a computer-aided diagnosis scheme for detection of lacunar infarcts in order to assist radiologists' interpretation as a "second opinion." We first segmented the cerebral region in the T1-weighted image by using a region growing technique. For identifying the initial candidates of lacunar infarcts, white top-hat transform and multiple-phase binarization were then applied to the T2-weighted image. 12 features, i.e., the locations, density differences from T1-and T2-weighted images, nodular components, and nodular & linear components were determined in the initial candidates regions. The rule-based schemes and an artificial neural network with 12 features were employed for distinguishing between lacunar infarcts and false positives. The sensitivity for detection of lacunar infarcts was 96.8% (90/93) with 0.69 FPs per image. Our computerized scheme might be useful in assisting radiologists for identifying lacunar infarcts in MR images.
社団法人電子情報通信学会の論文
2007-01-19