Non-destructive Testing (NDT) often involves analysing images to identify (rare) defects. We propose a method for locating and classifying abnormalities using Convolutional Neural Networks (CNNs). A ...particular problem is that it is often difficult to get large numbers of examples of images of defects, making training a classifier challenging. To address this problem we generate large numbers of synthetic images by combining real defects with different backgrounds. These images are used to train a U-Net style network to perform defect detection at the pixel level. We also demonstrate that the encoder of the network produces features which can be applied to the defect classification task at the image level. Both the defect detection and classification modules are tested on multiple small data sets. Our results show that these modules are able to fulfil the industrial component inspection task at the pixel level (locating defect regions) and image level (identifying if an image contains a defect).
A widely used approach for locating points on deformable objects in images is to generate feature response images for each point, and then to fit a shape model to these response images. We ...demonstrate that Random Forest regression-voting can be used to generate high quality response images quickly. Rather than using a generative or a discriminative model to evaluate each pixel, a regressor is used to cast votes for the optimal position of each point. We show that this leads to fast and accurate shape model matching when applied in the Constrained Local Model framework. We evaluate the technique in detail, and compare it with a range of commonly used alternatives across application areas: the annotation of the joints of the hands in radiographs and the detection of feature points in facial images. We show that our approach outperforms alternative techniques, achieving what we believe to be the most accurate results yet published for hand joint annotation and state-of-the-art performance for facial feature point detection.
•We organized two challenges for landmark detection, pathology classification and teeth segmentation in dental x-ray image analysis.•Datasets include 400 cephalometric images and 120 bitewing images ...with a referenced standard generated by medical experts.•The datasets and the evaluation software will be made available to the research community, further encouraging future developments in this field.
Display omitted
Dental radiography plays an important role in clinical diagnosis, treatment and surgery. In recent years, efforts have been made on developing computerized dental X-ray image analysis systems for clinical usages. A novel framework for objective evaluation of automatic dental radiography analysis algorithms has been established under the auspices of the IEEE International Symposium on Biomedical Imaging 2015 Bitewing Radiography Caries Detection Challenge and Cephalometric X-ray Image Analysis Challenge. In this article, we present the datasets, methods and results of the challenge and lay down the principles for future uses of this benchmark. The main contributions of the challenge include the creation of the dental anatomy data repository of bitewing radiographs, the creation of the anatomical abnormality classification data repository of cephalometric radiographs, and the definition of objective quantitative evaluation for comparison and ranking of the algorithms. With this benchmark, seven automatic methods for analysing cephalometric X-ray image and two automatic methods for detecting bitewing radiography caries have been compared, and detailed quantitative evaluation results are presented in this paper. Based on the quantitative evaluation results, we believe automatic dental radiography analysis is still a challenging and unsolved problem. The datasets and the evaluation software will be made available to the research community, further encouraging future developments in this field. (http://www-o.ntust.edu.tw/~cweiwang/ISBI2015/)
Automatic industrial inspection is critical to modern manufacturing enterprises. Due to their low cost and real-time processing speed, vision-based inspection systems are often used for this task. ...Deep Convolutional Neural Networks (CNNs) have been extensively applied to many computer vision tasks but require large numbers of annotated examples to train. Obtaining such annotations is expensive and time-consuming. In this study we describe a method which aims to make the best use of unannotated image data, which can often be collected easily. We propose a novel unsupervised local deep feature learning method based on image segmentation to build a network which can extract useful features from an image. The training algorithm alternates between (1) obtaining pseudo-labels by clustering the features extracted using a segmentation CNN and (2) training the CNN for feature learning using these pseudo-labels. To our knowledge, unsupervised local deep feature learning has not been addressed based on image segmentation in this way before. We demonstrate the approach on two aerospace weld inspection tasks. Our results show that the proposed unsupervised method performs almost as well as a CNN with the same architecture trained in a supervised manner.
•Introduction of an unsupervised local deep feature learning method.•Developing an automatic aerospace weld inspection system on top of the proposed method.•Comparing the proposed method with the unsupervised classification CNN with the same backbone.
Cephalometric tracing is a standard analysis tool for orthodontic diagnosis and treatment planning. The aim of this study was to develop and validate a fully automatic landmark annotation (FALA) ...system for finding cephalometric landmarks in lateral cephalograms and its application to the classification of skeletal malformations. Digital cephalograms of 400 subjects (age range: 7-76 years) were available. All cephalograms had been manually traced by two experienced orthodontists with 19 cephalometric landmarks, and eight clinical parameters had been calculated for each subject. A FALA system to locate the 19 landmarks in lateral cephalograms was developed. The system was evaluated via comparison to the manual tracings, and the automatically located landmarks were used for classification of the clinical parameters. The system achieved an average point-to-point error of 1.2 mm, and 84.7% of landmarks were located within the clinically accepted precision range of 2.0 mm. The automatic landmark localisation performance was within the inter-observer variability between two clinical experts. The automatic classification achieved an average classification accuracy of 83.4% which was comparable to an experienced orthodontist. The FALA system rapidly and accurately locates and analyses cephalometric landmarks in lateral cephalograms, and has the potential to significantly improve the clinical work flow in orthodontic treatment.
Background and purpose - Being able to predict the hip-knee-ankle angle (HKAA) from standard knee radiographs allows studies on malalignment in cohorts lacking full-limb radiography. We aimed to ...develop an automated image analysis pipeline to measure the femoro-tibial angle (FTA) from standard knee radiographs and test various FTA definitions to predict the HKAA.
Patients and methods - We included 110 pairs of standard knee and full-limb radiographs. Automatic search algorithms found anatomic landmarks on standard knee radiographs. Based on these landmarks, the FTA was automatically calculated according to 9 different definitions (6 described in the literature and 3 newly developed). Pearson and intra-class correlation coefficient ICC) were determined between the FTA and HKAA as measured on full-limb radiographs. Subsequently, the top 4 FTA definitions were used to predict the HKAA in a 5-fold cross-validation setting.
Results - Across all pairs of images, the Pearson correlations between FTA and HKAA ranged between 0.83 and 0.90. The ICC values from 0.83 to 0.90. In the cross-validation experiments to predict the HKAA, these values decreased only minimally. The mean absolute error for the best method to predict the HKAA from standard knee radiographs was 1.8° (SD 1.3).
Interpretation - We showed that the HKAA can be automatically predicted from standard knee radiographs with fair accuracy and high correlation compared with the true HKAA. Therefore, this method enables research of the relationship between malalignment and knee pathology in large (epidemiological) studies lacking full-limb radiography.
We introduce a new architecture called ChoiceNet where each layer of the network is highly connected with skip connections and channelwise concatenations. This enables the network to alleviate the ...problem of vanishing gradients, reduces the number of parameters without sacrificing performance and encourages feature reuse. We evaluate our proposed architecture on three independent tasks: classification, segmentation and facial landmark localisation. For this, we use benchmark datasets such as ImageNet, CIFAR-10, CIFAR-100, SVHN CamVid and 300W.
We describe a method for automatically building statistical shape models from a training set of example boundaries/surfaces. These models show considerable promise as a basis for segmenting and ...interpreting images. One of the drawbacks of the approach is, however, the need to establish a set of dense correspondences between all members of a set of training shapes. Often this is achieved by locating a set of "landmarks" manually on each training image, which is time consuming and subjective in two dimensions and almost impossible in three dimensions. We describe how shape models can be built automatically by posing the correspondence problem as one of finding the parameterization for each shape in the training set. We select the set of parameterizations that build the "best" model. We define "best" as that which minimizes the description length of the training set, arguing that this leads to models with good compactness, specificity and generalization ability. We show how a set of shape parameterizations can be represented and manipulated in order to build a minimum description length model. Results are given for several different training sets of two-dimensional boundaries, showing that the proposed method constructs better models than other approaches including manual landmarking-the current gold standard. We also show that the method can be extended straightforwardly to three dimensions.
Defect classification and detection have been explored using convolutional neural networks (CNNs). Normally, a large set of training images containing defects and the associated annotation data are ...required by these approaches. However, such a large set of images is usually difficult to collect because defects are rare and annotation is time-consuming and expensive. To address these issues, we propose to use a multitask deep one-class CNN for defect classification. Compared with supervised classification methods, this CNN does not require abnormal images and annotated data for training. Specifically, we build a stacked encoder-decoder autoencoder for learning feature representation from normal images. The encoder is used as a feature extractor based on the hard sharing scheme of multitask learning. A one-class classification (OCC) objective learned as a hypersphere using minimum volume estimation is appended to it. Together the encoder and the OCC objective lead to a deep one-class classifier. To train both the autoencoder and one-class classifier end-to-end, a multitask loss function is built. Given an unknown sample, the distance between its feature representation and the center of the hypersphere is used as the anomaly score. Furthermore, defect detection is implemented using a moving-window scanning method on top of the deep one-class classifier. The proposed approach achieves better performance than its counterparts trained using a two-stage method. For defect detection, our approach achieves results almost as good as the supervised method even without using any annotated data. We attribute the promising results to the advantages of multitask learning.