A two-stage approach for discriminating melanocytic skin lesions using standard cameras

In this paper, we propose a novel approach to discriminate malignant melanomas and benign atypical nevi, since both types of melanocytic skin lesions have very similar characteristics. Recent studies involving the non-invasive diagnosis of melanoma indicate that the concentrations of the two main classes of melanin present in the human skin, eumelanin and pheomelanin, can potentially be used in the computation of relevant features to differentiate these lesions. So, we describe how these features can be estimated using only standard camera images. Moreover, we demonstrate that using these features in conjunction with features based on the well known ABCD rule, it is possible to achieve 100% of sensitivity and more than 99% accuracy in melanocytic skin lesion discrimination, which is a highly desirable characteristic in a prescreening system.     

Toward a combined tool to assist dermatologists in melanoma detection from dermoscopic images of pigmented skin lesions

In this paper we propose a machine learning approach to classify melanocytic lesions as malignant or benign, using dermoscopic images. The lesion features used in the classification framework are inspired on border, texture, color and structures used in popular dermoscopy algorithms performed by clinicians by visual inspection. The main weakness of dermoscopy algorithms is the selection of a set of weights and thresholds, that appear not to be robust or independent of population. The use of machine learning techniques allows to overcome this issue. The proposed method is designed and tested on an image database composed of 655 images of melanocytic lesions: 544 benign lesions and 111 malignant melanoma. After an image pre-processing stage that includes hair removal filtering, each image is automatically segmented using well known image segmentation algorithms. Then, each lesion is characterized by a feature vector that contains shape, color and texture information, as well as local and global parameters. The detection of particular dermoscopic patterns associated with melanoma is also addressed, and its inclusion in the classification framework is discussed. The learning and classification stage is performed using AdaBoost with C4.5 decision trees. For the automatically segmented database, classification delivered a specificity of 77% for a sensitivity of 90%. The same classification procedure applied to images manually segmented by an experienced dermatologist yielded a specificity of 85% for a sensitivity of 90%.     

A narrow band graph partitioning method for skin lesion segmentation

Accurate skin lesion segmentation is critical for automated early skin cancer detection and diagnosis. In this paper, we present a novel multi-modal skin lesion segmentation method based on region fusion and narrow band energy graph partitioning. The proposed method can handle challenging characteristics of skin lesions, such as topological changes, weak or false edges, and asymmetry. Extensive testing demonstrated that in this method complex contours are detected correctly while topological changes of evolving curves are managed naturally. The accuracy of the method was quantified using a lesion similarity measure and lesion segmentation error ratio, Our results were validated using a large set of epiluminescence microscopy (ELM) images acquired using cross-polarization ELM and side-transillumination ELM. Our findings demonstrate that the new method can achieve improved robustness and better overall performance compared to other state-of-the-art segmentation methods.     

A practical approach for active camera coordination based on a fusion-driven multi-agent system

In this paper, we propose a multi-agent system architecture to manage spatially distributed active (or pan-tilt-zoom) cameras. Traditional video surveillance algorithms are of no use for active cameras, and we have to look at different approaches. Such multi-sensor surveillance systems have to be designed to solve two related problems: data fusion and coordinated sensor-task management. Generally, architectures proposed for the coordinated operation of multiple cameras are based on the centralisation of management decisions at the fusion centre. However, the existence of intelligent sensors capable of decision making brings with it the possibility of conceiving alternative decentralised architectures. This problem is approached by means of a MAS, integrating data fusion as an integral part of the architecture for distributed coordination purposes. This paper presents the MAS architecture and system agents.     

A computer-aided system for malignancy risk assessment of nodules in thyroid US images based on boundary features

In this paper, a novel computer-based approach is proposed for malignancy risk assessment of thyroid nodules in ultrasound images. The proposed approach is based on boundary features and is motivated by the correlation which has been addressed in medical literature between nodule boundary irregularity and malignancy risk. In addition, local echogenicity variance is utilized so as to incorporate information associated with local echogenicity distribution within nodule boundary neighborhood. Such information is valuable for the discrimination of high-risk nodules with blurred boundaries from medium-risk nodules with regular boundaries. Analysis of variance is performed, indicating that each boundary feature under study provides statistically significant information for the discrimination of thyroid nodules in ultrasound images, in terms of malignancy risk. k-nearest neighbor and support vector machine classifiers are employed for the classification tasks, utilizing feature vectors derived from all combinations of features under study. The classification results are evaluated with the use of the receiver operating characteristic. It is derived that the proposed approach is capable of discriminating between medium-risk and high-risk nodules, obtaining an area under curve, which reaches 0.95.     

A disk expansion segmentation method for ultrasonic breast lesions

Automatically extracting lesion boundaries in ultrasound images is difficult due to the variance in shape and interference from speckle noise. An effective scheme of removing speckle noise can facilitate the segmentation of ultrasonic breast lesions, which can be performed with an iterative disk expansion method. In this study, a disk expansion segmentation method is proposed to semi-automatically find lesion contours in ultrasonic breast image. To evaluate the performance of the proposed method, the simulations with seven types of cysts, three in vitro phantom images and 10 clinical breast images are introduced. The mean normalized true positive area overlap between simulated contours and contours obtained by the proposed method is over 85% in simulation results. A strong correlation exists between physicians’ manual delineations and detected contours in clinical breast images. In addition, the method is also verified to be able to simultaneously contour multiple lesions in a single image. In comparison with the conventional active contour model, our proposed method does not require any initial seed within a lesion and thus, it is more convenient and applicable.     

Two graph theory based methods for identifying the pectoral muscle in mammograms

Two image segmentation methods based on graph theory are used in conjunction with active contours to segment the pectoral muscle in screening mammograms. One method is based on adaptive pyramids (AP) and the other is based on minimum spanning trees (MST). The algorithms are tested on a public data set of mammograms and results are compared with previously reported methods. In 80% of the images, the boundary of the segmented regions has average error less than 2 mm. In 82 of 84 images, the boundary of the pectoral muscle found by the AP algorithm has average error less than 5 mm.     

A robust incremental learning framework for accurate skin region segmentation in color images

In this paper, we propose a robust incremental learning framework for accurate skin region segmentation in real-life images. The proposed framework is able to automatically learn the skin color information from each test image in real-time and generate the specific skin model (SSM) for that image. Consequently, the SSM can adapt to a certain image, in which the skin colors may vary from one region to another due to illumination conditions and inherent skin colors. The proposed framework consists of multiple iterations to learn the SSM, and each iteration comprises two major steps: (1) collecting new skin samples by region growing; (2) updating the skin model incrementally with the available skin samples. After the skin model converges (i.e., becomes the SSM), a post-processing can be further performed to fill up the interstices on the skin map. We performed a set of experiments on a large-scale real-life image database and our method observably outperformed the well-known Bayesian histogram. The experimental results confirm that the SSM is more robust than static skin models.     

Segmentation of malignant tumours in mammogram images: A hybrid approach using convolutional neural networks and connected component analysis

The segmentation of breast lesions is an important step in the computer-aided analysis of the mammogram. The presence of noise in mammograms makes lesion detection challenging particularly for complex malignant lesions. Pre-processing techniques can deal with the noise issue but distorts the important shape features. This motivates us to propose a novel hybrid approach by combining a convolution neural network (CNN) with connected component analysis (CCA) to segment malignant breast lesions without any pre-processing to avoid any distortion in image sharpness at the initial stages. Two well-known segmentation techniques namely, K-means (KM) and Fuzzy c-means (FCM) are also used to compare the results. From a pool of 1045 mammographic cancer images acquired from the Digital Database for Screening Mammography (DDSM), 1016 are used for training and validation, and 29 are used for testing. All three results (Hybrid, KM and FCM) are compared against the results by the expert Radiologist. The results indicate that, among various segmentation techniques, the proposed hybrid approach achieves the highest accuracy (90%), Matthew's correlation coefficient (0.79), Jaccard index (0.73) and the Dice similarity coefficient (0.84). Other performance evaluation techniques such as; precision, sensitivity, specificity, false-positive rate, false discovery rate, negative predictive value and false-negative rate also show the superior performance of the proposed hybrid approach. Statistical analysis (Mann–Whitney U test, T-test, Chi-square test, Kolmogorov–Smirnov test and Wilcoxon test), graphical analysis (Regression and Bland–Altman plots) and receiver operating characteristic curve further demonstrate the stability and consistency of the results.     

A novel approach for tracking high speed skaters in sports using a panning camera

This paper presents a computer vision system for tracking high-speed non-rigid skaters over a larger rink in short track speed skating competitions. The outputs of the tracking system are spatio-temporal trajectories of the skaters which can be further processed and analyzed by sports experts. To capture highly complex and dynamic scenes, the camera pans very fast, therefore, tracking amorphous skaters becomes a challenging task. We propose a new method for (1) automatically computing the transformation matrices to map each frame to the globally consistent model of the rink; (2) incorporating the hierarchical model based on the contextual knowledge and multiple cues into the unscented Kalman filter to improve the tracking performance when occlusions occur; (3) evaluating the precision of our practical system objectively. Experimental results show that the proposed algorithm is very efficient and effective on the video recorded in the World Short Track Speed Skating Championships.