Abstract
Computed tomography (CT) is used widely to image pa- tients for medical diagnosis and to scan baggage for threatening materi- als. Automated reading of these images can be used to reduce the costs of a human operator, extract quantitative information from the images or support the judgements of a human operator. Ob ject quantification requires an image segmentation to make measurements about ob ject size, material composition and morphology. Medical applications mostly re- quire the segmentation of prespecified ob jects, such as specific organs or lesions, which allows the use of customized algorithms that take ad- vantage of training data to provide orientation and anatomical context of the segmentation targets. In contrast, baggage screening requires the segmentation algorithm to provide segmentation of an unspecified num- ber of ob jects with enormous variability in size, shape, appearance and spatial context. Furthermore, security systems demand 3D segmentation algorithms that can quickly and reliably detect threats. To address this problem, we present a segmentation algorithm for 3D CT images that makes no assumptions on the number of ob jects in the image or on the composition of these ob jects. The algorithm features a new Automatic QUality Measure (AQUA) model that measures the segmentation confi- dence for any single ob ject (from any segmentation method) and uses this confidence measure to both control splitting and to optimize the segmen- tation parameters at runtime for each dataset. The algorithm is tested on 27 bags that were packed with a large variety of different ob jects.