We present an efficient neural network method for locating anatomical landmarks in 3D medical CT scans, using atlas location autocontext in order to learn long-range spatial context. Location predictions are made by regression to Gaussian heatmaps, one heatmap per landmark. This system allows patchwise application of a shallow network, thus enabling multiple volumetric heatmaps to be predicted concurrently without prohibitive GPU memory requirements. Further, the system allows inter-landmark spatial relationships to be exploited using a simple overdetermined affine mapping that is robust to detection failures and occlusion or partial views. Evaluation is performed for 22 landmarks defined on a range of structures in head CT scans. Models are trained and validated on 201 scans. Over the final test set of 20 scans which was independently annotated by 2 human annotators, the neural network reaches an accuracy which matches the annotator variability, with similar human and machine patterns of variability across landmark classes.

The recommended exam for assessing chest trauma is a computed tomography (CT) chest scan. Using multi-planar reconstructions to evaluate a CT volume to assess the ribcage is a tedious and time-consuming task. We have designed an application that provides an automatically rendered unfolded unobstructed view of the entire ribcage using an unfolded cylindrical projection. This paper describes the underlying algorithm which has two main steps: ribcage segmentation and ribcage unfolding. The unfolding technique we developed preserves the relative size and location of the ribs and surrounding tissue, providing a natural anatomical reference for the reader. It also demonstrated usefulness to identify other musculoskeletal conditions such us scoliosis, calcified cartilage, bone tumours. To evaluate the usefulness of the application, we evaluated it on 70 representative CT chest scans. The evaluation was performed by a clinical expert who graded the specialized unfolded cylindrical projection view on a 5 point Likert scale according to the level of diagnostic confidence. Results showed that 84% of the studies were clinically useful (above grade 3). The algorithm is fully automatic and it runs in an average time of 24 s. The evaluation described in this paper gives positive initial feedback on the usefulness of the application. A recent multi-reader clinical study showed that using the specialized unfolded cylindrical projection view obtains similar diagnostic accuracy to conventional multi-planar reconstructions while reducing the reading time.

We present an efficient neural network approach for locating anatomical landmarks, using a two-pass, two-resolution cascaded approach which leverages a mechanism we term atlas location autocontext. Location predictions are made by regression to Gaussian heatmaps, one heatmap per landmark. This system allows patchwise application of a shallow network, thus enabling the prediction of multiple volumetric heatmaps in a unified system, without prohibitive GPU memory requirements. Evaluation is performed for 22 landmarks defined on a range of structures in head CT scans and the neural network model is benchmarked against a previously reported decision forest model trained with the same cascaded system. Models are trained and validated on 201 scans. Over the final test set of 20 scans which was independently annotated by 2 observers, we show that the neural network reaches an accuracy which matches the annotator variability.

Variations in the shape and appearance of anatomical structures in medical images are often relevant radiological signs of disease. Automatic tools can help automate parts of this manual process. A cloud-based evaluation framework is presented in this paper including results of benchmarking current state-of-the-art medical imaging algorithms for anatomical structure segmentation and landmark detection: the VISCERAL Anatomy benchmarks. The algorithms are implemented in virtual machines in the cloud where participants can only access the training data and can be run privately by the benchmark administrators to objectively compare their performance in an unseen common test set. Overall, 120 computed tomography and magnetic resonance patient volumes were manually annotated to create a standard Gold Corpus containing a total of 1295 structures and 1760 landmarks. Ten participants contributed with automatic algorithms for the organ segmentation task, and three for the landmark localization task. Different algorithms obtained the best scores in the four available imaging modalities and for subsets of anatomical structures. The annotation framework, resulting data set, evaluation setup, results and performance analysis from the three VISCERAL Anatomy benchmarks are presented in this article. Both the VISCERAL data set and Silver Corpus generated with the fusion of the participant algorithms on a larger set of non-manually-annotated medical images are available to the research community.

The automatic detection and localization of anatomical landmarks has wide application, including intra and interpatient registration, study location and navigation, and the targeting of specialized algorithms. In this paper, we demonstrate the automatic detection and localization of 127 anatomically defined landmarks distributed throughout the body, excluding arms. Landmarks are defined on the skeleton, vasculature and major organs. Our approach builds on the classification forests method,1 using this classifier with simple image features which can be efficiently computed. For the training and validation of the method we have used 369 CT volumes on which radiographers and anatomists have marked ground truth (GT) - that is the locations of all defined landmarks occurring in that volume. A particular challenge is to deal with the wide diversity of datasets encountered in radiology practice. These include data from all major scanner manufacturers, different extents covering single and multiple body compartments, truncated cardiac acquisitions, with and without contrast. Cases with stents and catheters are also represented. Validation is by a leave-one-out method, which we show can be efficiently implemented in the context of decision forest methods. Mean location accuracy of detected landmarks is 13.45mm overall; execution time averages 7s per volume on a modern server machine. We also present localization ROC analysis to characterize detection accuracy - that is to decide if a landmark is or is not present in a given dataset.