一种基于水平集的胸部淋巴结全自动分割算法

Automatic Chest Lymph Node Segmentation Based on Level Set Approach

肺癌是世界上发病率和死亡率最高的恶性肿瘤。现阶段,临床主要通过淋巴结病变程度判断肺癌恶性程度和预后情况。针对淋巴结的解剖特征及全身分散的问题,本文提出了一种全自动检测和分割胸部病变淋巴结的方法,首先利用全自动解剖识别方法自动识别定位PET-CT影像中的所有淋巴结区域,然后检测每个淋巴结区域中潜在的病变淋巴结,第三步采用水平集模型完成对各个淋巴结区域内部病变淋巴结的精准分割。算法分割精度率平均达到85%。

Lymph node detection is challenging due to the low contrast between lymph nodes as well as surrounding soft tissues and the variation in nodal size and shape.Currently,the degree of malignancy and prognosis of lung cancer are mainly determined by the degree of lymph node lesions.Aiming at the anatomical features of lymph nodes and the problem of systemic dispersal,this paper proposes a method for fully automatic detection and segmentation of thoracic lymph nodes.Firstly,the automatic anatomical recognition method is used to automatically identify and locate all lymph node regions in PET-CT images,and then detect each.The potential lymph nodes in the lymph node area are detected.The third step uses the level set model to complete segmentation of the diseased lymph nodes in each lymph node region.The algorithm segmentation accuracy rate averages 85%.