We introduce a robust multi-object segmentation algorithm based on visual patch classification for abdominal CT segmentation. Firstly, the proximity of the pixels is expressed by both intensity and spatial distance. And then clustering framework is employed to form various visual patches. In this way, the noise and embedded small tissues such as blood vessels and tracheas which often make other segmentation algorithms failed are filtered out during the cluster iteration. Afterwards, the visual patches are further grouped by the way of classification in the criteria of spatial relationship of visual pitches. Specially, the algorithm can be viewed as effectively tradeoff of bottom-up methods and top-down methods. The approach has been applied to the multi-object segmentation of abdominal CT images, such as the liver, kidney, spleen and gallbladder. We have test the method in American published TCIA database, whose efficiency and robustness is evaluated through quantification results on both sectional level and volumetric level, which exhibit the optimistic application and prospect in the field of medical image processing.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
- Multi-object Segmentation for Abdominal CT Image Based on Visual Patch Classification
- Springer Berlin Heidelberg
Neuer Inhalt/© ITandMEDIA