Ovarian tumors are the most common diagnostic challenge for gynecologists and ultrasound examination has become the main technique for assessment of ovarian pathology and for preoperative distinction between malignant and benign ovarian tumors. However, ultrasonography is highly examiner-dependent and there may be an important variability between two different specialists when examining the same case. The objective of this work is the evaluation of different well-known Machine Learning (ML) systems to perform the automatic categorization of ovarian tumors from ultrasound images. We have used a real patient database whose input features have been extracted from 348 images, from the IOTA tumor images database, holding together with the class labels of the images. For each patient case and ultrasound image, its input features have been previously extracted using Fourier descriptors computed on the Region Of Interest (ROI). Then, four ML techniques are considered for performing the classification stage: K-Nearest Neighbors (KNN), Linear Discriminant (LD), Support Vector Machine (SVM) and Extreme Learning Machine (ELM). According to our obtained results, the KNN classifier provides inaccurate predictions (less than 60% of accuracy) independently of the size of the local approximation, whereas the classifiers based on LD, SVM and ELM are robust in this biomedical classification (more than 85% of accuracy). ML methods can be efficiently used for developing the classification stage in computer-aided diagnosis systems of ovarian tumor from ultrasound images. These approaches are able to provide automatic classification with a high rate of accuracy. Future work should aim at enhancing the classifier design using ensemble techniques. Another ongoing work is to exploit different kind of features extracted from ultrasound images.
22 September 2023 in Medical & Biological Engineering & Computing
Deep convolutional neural networks (DCNNs) have demonstrated promising performance in classifying breast lesions in 2D ultrasound (US) images. Exiting approaches typically use pre-trained models based on architectures designed for natural images with transfer learning. Fewer attempts have been made to design customized architectures specifically for this purpose. This paper presents a comprehensive evaluation on transfer learning based solutions and automatically designed networks, analyzing the accuracy and robustness of different recognition models in three folds. First, we develop six different DCNN models (BNet, GNet, SqNet, DsNet, RsNet, IncReNet) based on transfer learning. Second, we adapt the Bayesian optimization method to optimize a CNN network (BONet) for classifying breast lesions. A retrospective dataset of 3034 US images collected from various hospitals is then used for evaluation. Extensive tests show that the BONet outperforms other models, exhibiting higher accuracy (83.33%), lower generalization gap (1.85%), shorter training time (66 min), and less model complexity (approximately 0.5 million weight parameters). We also compare the diagnostic performance of all models against that by three experienced radiologists. Finally, we explore the use of saliency maps to explain the classification decisions made by different models. Our investigation shows that saliency maps can assist in comprehending the classification decisions. Graphical Abstract
1 March 2017 in Computer Methods and Programs in Biomedicine
Data stream clustering is an active area of research in big data. It refers to clustering constantly arriving new data records and updating existing cluster patterns and outliers in light of the newly arriving data. Density-based algorithms for solving this problem have the promise for finding arbitrary shape clusters and detecting anomalies without prior knowledge of the number of clusters. In this paper, a new incremental algorithm known as Enhanced Density-based Data Stream (EDDS) is developed to overcome limitations with the existing solutions. The algorithm detects clusters and outliers in an incoming data chunk, merges new clusters from the chunk with the existing clusters, and filters out new outliers for the next round. It modified the traditional DBSCAN algorithm to summarise each cluster in terms of a set of surface-core points. The algorithm applies the density-reachable concept of DBSCAN as its merging strategy and prunes the internal core points using a heuristic solution. The algorithm also removes the aged core points and outliers depending on a fading function. The paper investigates three versions of the algorithm for three possible representations of clusters where either all core points are maintained (EDDS-I), only core points of the new clusters from the incoming chunk are kept (EDDS-II), or only the surface-core points of the cluster shapes are kept (EDDS-III) to examine the balance between the efficiency gain for the algorithm and the amount of overhead time committed for pruning internal core points. The algorithm was evaluated on selected datasets using various quality measures. The experimental results indicate improvements in terms of clustering correctness with a comparable time complexity over the existing solutions for solving the same kind of problems.