Spline-Based Probabilistic Model for Anatomical Landmark Detection

Izard, Camille; Jedynak, Bruno; Stark, Craig E.L.

doi:10.1007/11866565_104

Cited by 12 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare the performance obtained with this joint algorithm with that obtained with the simplified version introduced in Izard et al (2006). The simplified model essentially decouples the estimation of the photometry and the geometry in the learning and in the testing algorithms.…”

Section: Template Estimationmentioning

confidence: 97%

“…The red curve represents the evolution of the likelihood by joint optimization. The blue curve represents the likelihood evolution when using the decoupled algorithm and finally the green curve represents the evolution of the likelihood when using the joint algorithm, initializing with the template estimate given by the decoupled algorithm presented in Izard et al (2006). The experiment was performed around SCC1, using 30 images for training, modeling two tissue types.…”

Section: Template Estimationmentioning

confidence: 99%

“…To assess the advantage of the joint optimization compared to the decoupled optimization in terms of detection, we performed 4 experiments. In the first experiment, denoted by DD, we use the decoupled algorithm detailed in Izard et al (2006) to perform the detection. In the second experiment, denoted by JD, we use the joint estimation to select the model parameters but perform the landmark detection using the decoupled algorithm.…”

Section: Detection Performancementioning

confidence: 99%

See 2 more Smart Citations

Learning to Match: Deriving Optimal Template-Matching Algorithms from Probabilistic Image Models

Vidal

Jedynak

2009

Int J Comput Vis

Self Cite

View full text Add to dashboard Cite

Finding correspondences between images by template matching is a common problem in image understanding. Although a variety of solutions have been proposed, most of them rely on the arbitrary choice of a template and a matching function. Often, different cost functions lead to different results, and the choice of a good cost for a specific application remains an art. Statistical models on the other hand, allow us to derive optimal learning and matching algorithms from modeling assumptions using likelihood maximization principles. The key contribution of this paper is the development of a statistical framework for learning what function to optimize from training examples. We present a family of statistical models for grayscale images, which allow us to derive optimal template-matching algorithms. The intensity at each pixel is described by a random variable whose distribution is encoded by a deformable template. Firstly, we assume the intensity distribution to be Gaussian and derive an intensity-matching algorithm, which is a generalization of the classical sum-of-squared differences. Then, we introduce a hidden segmentation variable in the probabilistic model and derive a segmentation-matching algorithm that can handle photometric variations. Both models are exemplified on the automatic detection of anatomical landmarks in brain Magnetic Resonance Images.

show abstract

Section: Template Estimationmentioning

confidence: 97%

Section: Template Estimationmentioning

confidence: 99%

Section: Detection Performancementioning

confidence: 99%

See 1 more Smart Citation

Learning to Match: Deriving Optimal Template-Matching Algorithms from Probabilistic Image Models

Vidal

Jedynak

2009

Int J Comput Vis

Self Cite

View full text Add to dashboard Cite

show abstract

“…Pauly et al [3] simultaneously regress out the locations and sizes of multiple organs with confidence scores using a learned Random Forest regressor. To some extent, image registration [11] can be regarded as using the holistic context too.…”

Section: Related Work and Context Exploitationmentioning

confidence: 99%

Anatomical Landmark Detection Using Nearest Neighbor Matching and Submodular Optimization

Liu

Zhou

2012

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2012

View full text Add to dashboard Cite

Abstract. We present a two-stage method for effective and efficient detection of one or multiple anatomical landmarks in an arbitrary 3D volume. The first stage of nearest neighbor matching is to roughly estimate the landmark locations. It searches out of 100,000 volumes for the closest to an input volume and then transfers landmark annotations to the input. The second stage of submodular optimization is to refine the landmark locations by running discriminative landmark detectors within the search ranges constrained by the first stage results. Further it coordinates multiple detectors with a search strategy optimized on the fly to reduce the overall computation cost arising in a submodular formulation. We validate the accuracy, speed and robustness of our approach by detecting body regions and landmarks in a dataset of 2,500 CT scans.

show abstract

“…Furthermore, such methods aren't able to locate individual anatomical landmarks. The atlas method [3,33] is also used to extract anatomical landmarks. Disadvantages of this method are: (i) the atlas has to be given beforehand; (ii) when the atlas is corrected after it is mapped to a new surface, geometrical variables of the new surface still have to be evaluated.…”

Section: Introductionmentioning

confidence: 99%

Automatic extraction of 3D anatomical feature curves of hip bone models reconstructed from CT images

Liu

Qian

Zhao

2015

BME

View full text Add to dashboard Cite

Abstract. 3D anatomical feature curves (AFC) on bone models reconstructed from CT/MRI images are important in some fields, such as preoperative planning, intra-operative navigation, patient-specific prosthesis design, etc. Interactive extraction of feature curves on patient-specific bone models is time-consuming, has low repeatability and accuracy. This paper presents a computer graphics method to automatically extract AFC from 3D hip bone models reconstructed from CT images. A DCSS (direct curvature scale space)-based technique is firstly used to extract anatomical feature points (AFP) in every contour, using anatomical structure information as prior knowledge so that AFP are extracted and only extracted. Then, corresponding AFP are linked in different contours and AFC is generated. AFC obtained by our method were compared with those interactively extracted by three surgeons, which showed that our method is feasible (Dice coefficient: 0.94; Average symmetric surface distance: 3.97 mm). The method was also applied to identify anatomical landmarks, which showed that our method is superior to the curvature-based methods that fail to identify landmark regions or have too many redundant regions, which results in failures to subsequently label landmark regions using pre-defined spatial adjacency matrices.

show abstract

Spline-Based Probabilistic Model for Anatomical Landmark Detection

Cited by 12 publications

References 15 publications

Learning to Match: Deriving Optimal Template-Matching Algorithms from Probabilistic Image Models

Learning to Match: Deriving Optimal Template-Matching Algorithms from Probabilistic Image Models

Anatomical Landmark Detection Using Nearest Neighbor Matching and Submodular Optimization

Automatic extraction of 3D anatomical feature curves of hip bone models reconstructed from CT images

Contact Info

Product

Resources

About