| 摘要 |
PURPOSE: An apparatus and method for diagnosing a mucinous cystic tumor are provided to automatically classify a plurality of inner cavity regions from given tissue cell images using a direction cumulative histogram. CONSTITUTION: A control part of apparatus for diagnosing a mucinous cystic tumor receives tissue cell images from an image input part(S110). The control part identifies an inner cavity region from the tissue cell images by controlling an inner cavity region identity part(S120). The control part identifies an epithelial cell by controlling an epithelial cell identification part(S130). The control part extracts existing morphological characteristics and new characteristics specialized in a cystic tumor from nucleuses of extracted epithelial cell by controlling a characteristic extraction part(S140). A diagnostic part detects the mucinous cystic tumor from the tissue cell images using a SVM(Support Vector Machine) which learns the extracted characteristics(S150). [Reference numerals] (S110) Tissue cell image input; (S120) Inner cavity region identification; (S121) Preprocessing; (S122) Obtaining of a candidate point for identifying the inner cavity region; (S123) Detection of an inner cavity boundary; (S130) Epithelial cell identification; (S131) Preprocessing; (S132) Identifying of a nucleus inside an image; (S133) Identifying of the nucleus inside the epithelial cell; (S140) Characteristics extraction; (S150) Diagnosis using an SVM model |
