医学图像感兴趣区域的自动提取

Automatic Extraction of ROI in Medical Image

针对医学图像归档与通信系统通过视频采集和胶片扫描产生的海量图像数据,提出了一种感兴趣区域自动提取和归一化的预处理算法。该算法具有时间复杂度低,无须人工干预,不会对提取的感兴趣区域图像数据造成任何损耗等特点,适合海量医学图像数据库的批量处理。将其应用于基于图像内容的归档与查询算法中,作为图像提取特征前的预处理算法,实验数据证明较传统直接针对原始采集图像,可有效改善特征提取准确度,提高查询精度。

A pretreatment algorithm of automatic extraction of ROI(Region Of Interested) is presented,which aims at daily data generated from Picture Archiving and Communication System(PACS), especially those video-captured and film-scanned images. It can be no harm to the ROI data, less time-consuming and no manual-intervened, which befits the batch pretreatment of magnanimity images database. The experiment prove that this algorithm can effect in raising the performance of the subsequent program in content-based images query.