Extraction of pulmonary fissures from thin‐section CT images by two‐dimensional linear feature detector method

We describe an algorithm for the extraction of interlobar fissures which is useful in discriminative diagnosis from thin-section CT images. In clinical diagnosis, interlobar fissures are important in identification of the lung structure and in discriminative diagnosis of lesions. Interlobar fissures are shadows whose density values are very low and whose shape differs greatly from that of the pulmonary blood vessels. The proposed technique performs reduction processing of linear artifacts on images and emphasis processing of interlobar fissures. Improved VanderBrug linear detection is used in this processing. Interlobar fissures are detected from the difference in structure by the morphology operations of erosion and dilation. The proposed technique is little affected by image artifacts. Its effectiveness is demonstrated by applying it to clinical images and evaluating the accuracy of the results. © 2005 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 88(10): 59–68, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20225