Learning When Negative Examples Abound
ECML '97 Proceedings of the 9th European Conference on Machine Learning
Digital Image Processing Using MATLAB
Digital Image Processing Using MATLAB
Histograms of Oriented Gradients for Human Detection
CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 1 - Volume 01
A Performance Evaluation of Local Descriptors
IEEE Transactions on Pattern Analysis and Machine Intelligence
DAISY: An Efficient Dense Descriptor Applied to Wide-Baseline Stereo
IEEE Transactions on Pattern Analysis and Machine Intelligence
SURF: speeded up robust features
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part I
Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering
IEEE Transactions on Image Processing
Towards segmentation of pedicles on posteroanterior x-ray views of scoliotic patients
ICIAR'07 Proceedings of the 4th international conference on Image Analysis and Recognition
| Hi-index | 0.00 |
Assessing spinal deformations requires a 3D evaluation. However, due to restrictions of conventional 3D imaging techniques, 3D reconstructions are typically performed from planar radiographs. Conventional reconstruction methods require a large interaction time for the identification of anatomical structures of interest. Recently, semi-supervised methods were proposed that enable to reduce interaction time. However, these methods have shown difficulties to determine precisely the pedicles of vertebrae, which are fundamental for calculating several clinical indices. This paper proposes a new method for the detection of pedicles in planar radiographs. The method is based in the use of feature descriptors for training a binary classifier and a detection phase that is carried out by sweeping a region of interest classifying all of its pixels. The location of the pedicle corresponds to the candidate with the largest output value of the classifier. The evaluation of the method was performed by comparison with a manual identification from an expert. The classifier used was a Support Vector Machine (SVM) and several descriptors were selected in order to determine which best suits this problem. The best results were obtained using Histograms of Oriented Gradients (HOG), which was able of determining a valid detection in approximatly half of the cases.