No abstract

In this paper, we propose a data balancing method for multi-label biomedical data. The method can be applied in the case of semantic segmentation problems for balancing the corresponding image data. The proposed method performs oversampling of instances of minority classes in a way that increases the frequencies of appearance (a ratio of number of samples, containing this class, over the total number of samples in the dataset) of minority classes in the data, thereby reducing the class imbalance. The effectiveness of the proposed method is shown experimentally by applying it to two highly unbalanced biomedical image datasets. A convolutional neural network (CNN) was trained on several versions of those datasets: one balanced with the proposed method, another balanced with manual oversampling and an unbalanced version. The results of the experiments validate the effectiveness of the proposed method, proving that it allows the influence of class imbalance on the learning algorithm to be reduced, thus improving its original classification results for most of the classes. Apart from biomedical image data, the proposed method was applied to several common multi-label datasets. Inherently, the proposed method does not make any assumptions about the underlying structure of the data to be balanced; therefore, it can be applied to all types of data (vectors, images, etc.) that can be described in a multi-label framework. It also can be used in conjunction with any learning algorithm that is suitable for multi-label data. To illustrate its wider applicability, a series of experiments was conducted using seven common multi-label datasets. An experimental comparison to existing multi-label data balancing approaches is provided, as well. The experimental results show that the proposed method presents a competitive alternative to existing approaches.

Advances in the neural networks have brought revolution in many areas, especially those related to image processing and analysis. The most complex is a task of analyzing biomedical data due to a limited number of samples, imbalanced classes, and low-quality labelling. In this paper, we look into the possibility of using neural networks when solving a task of semantic segmentation of fundus. The applicability of the neural networks is evaluated through a comparison of image segmentation results with those obtained using textural features. The neural networks are found to be more accurate than the textural features both in terms of precision (~25%) and recall (~50%). Neural networks can be applied in biomedical image segmentation in combination with data balancing algorithms and data augmentation techniques.

The paper is about the development of an approach which able to create rules for distinguish-ing between specified objects of hyperspectral data using a small number of observations. Such an approach would contribute to the development of methods and algorithms for the operational analysis of hyperspectral data. These methods can be used for hyperspectral data preprocessing and labeling. Implementation of the proposed approach are using a technology that harnesses both discriminative criteria and the general formulas of spectral indexes. In implementing the proposed technology, the index was defined with normalized difference formula. The Informativeness was estimated using separability criteria of discriminative analysis. The results show that the implemented algorithm can recognize areas of hyperspectral data with different vegetation. The index formed by the algorithm is similar to Normalized Difference Vegetation Index (NDVI). The proposed technology is the generalization of the approach of forming recognition rules using a small number of features. It has been shown that technology can form informative indexes in specified tasks of hyperspectral data analysis.

No abstract