Automatic anatomy recognition in whole-body PET/CT images

Medical Imaging 2018: Image Processing

et al. 2018

Self Cite

The recently developed body-wide Automatic Anatomy Recognition (AAR) methodology depends on fuzzy modeling of individual objects, hierarchically arranging objects, constructing an anatomy ensemble of these models, and a dichotomous object recognition–delineation process. The parent-to-offspring spatial relationship in the object hierarchy is crucial in the AAR method. We have found this relationship to be quite complex, and as such any improvement in capturing this relationship information in the anatomy model will improve the process of recognition itself. Currently, the method encodes this relationship based on the layout of the geometric centers of the objects. Motivated by the concept of virtual landmarks (VLs), this paper presents a new one-shot AAR recognition method that utilizes the VLs to learn object relationships by training a neural network to predict the pose and the VLs of an offspring object given the VLs of the parent object in the hierarchy. We set up two neural networks for each parent-offspring object pair in a body region, one for predicting the VLs and another for predicting the pose parameters. The VL-based learning/prediction method is evaluated on two object hierarchies involving 14 objects. We utilize 54 computed tomography (CT) image data sets of head and neck cancer patients and the associated object contours drawn by dosimetrists for routine radiation therapy treatment planning. The VL neural network method is found to yield more accurate object localization than the currently used simple AAR method.

Section: Methodsmentioning

confidence: 99%

Automatic anatomy recognition using neural network learning of object relationships via virtual landmarks

Yan

Medical Imaging 2018: Image Processing

et al. 2018

Self Cite

“…In [6], it builds a super mask with the union of all training binary images of the object and makes use of the statistical information of histograms of the gray image over super mask to estimate the optimum threshold. However, the range of threshold may sometimes shrink into a singular value for some organs like stomach.…”

Section: Methodsmentioning

confidence: 99%

“…Select an initial range of threshold by using the method in [6]. Determine the start value of the threshold S, the end value of the threshold E, the tolerance range of the threshold of the object from T1 to T2, and the value of step in each iteration;…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Thoracic lymph node station recognition on CT images based on automatic anatomy recognition with an optimal parent strategy

Medical Imaging 2018: Image Processing

et al. 2018

Self Cite

Currently, there are many papers that have been published on the detection and segmentation of lymph nodes from medical images. However, it is still a challenging problem owing to low contrast with surrounding soft tissues and the variations of lymph node size and shape on computed tomography (CT) images. This is particularly very difficult on low-dose CT of PET/CT acquisitions. In this study, we utilize our previous automatic anatomy recognition (AAR) framework to recognize the thoracic-lymph node stations defined by the International Association for the Study of Lung Cancer (IASLC) lymph node map. The lymph node stations themselves are viewed as anatomic objects and are localized by using a one-shot method in the AAR framework. Two strategies have been taken in this paper for integration into AAR framework. The first is to combine some lymph node stations into composite lymph node stations according to their geometrical nearness. The other is to find the optimal parent (organ or union of organs) as an anchor for each lymph node station based on the recognition error and thereby find an overall optimal hierarchy to arrange anchor organs and lymph node stations. Based on 28 contrast-enhanced thoracic CT image data sets for model building, 12 independent data sets for testing, our results show that thoracic lymph node stations can be localized within 2–3 voxels compared to the ground truth.

“…The method adapts to this application a recently developed technology called automatic anatomy recognition (AAR) [14,15] which aims to recognize and delineate automatically numerous internal solid organs body-wide on CT, MRI, and positron emission tomography (PET) images based on population fuzzy anatomy models.…”

Section: Introductionmentioning

confidence: 99%

Fat segmentation on chest CT images via fuzzy models

Medical Imaging 2016: Image-Guided Procedures, Robotic Interventions, and Modeling

et al. 2016

Self Cite

Quantification of fat throughout the body is vital for the study of many diseases. In the thorax, it is important for lung transplant candidates since obesity and being underweight are contraindications to lung transplantation given their associations with increased mortality. Common approaches for thoracic fat segmentation are all interactive in nature, requiring significant manual effort to draw the interfaces between fat and muscle with low efficiency and questionable repeatability. The goal of this paper is to explore a practical way for the segmentation of subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) components of chest fat based on a recently developed body-wide automatic anatomy recognition (AAR) methodology. The AAR approach involves 3 main steps: building a fuzzy anatomy model of the body region involving all its major representative objects, recognizing objects in any given test image, and delineating the objects. We made several modifications to these steps to develop an effective solution to delineate SAT/VAT components of fat. Two new objects representing interfaces of SAT and VAT regions with other tissues, SatIn and VatIn are defined, rather than using directly the SAT and VAT components as objects for constructing the models. A hierarchical arrangement of these new and other reference objects is built to facilitate their recognition in the hierarchical order. Subsequently, accurate delineations of the SAT/VAT components are derived from these objects. Unenhanced CT images from 40 lung transplant candidates were utilized in experimentally evaluating this new strategy. Mean object location error achieved was about 2 voxels and delineation error in terms of false positive and false negative volume fractions were, respectively, 0.07 and 0.1 for SAT and 0.04 and 0.2 for VAT.