Search

Machine learning (ML) is commonly used to develop survival-predictive radiomic models in non-small cell lung cancer (NSCLC) patients, which helps assist treatment decision making. Radiomic features derived from computer tomography (CT) lung images aim to capture quantitative tumor characteristics. However, these features are determined by humans, which poses a risk of including irrelevant or redundant variables, thus reducing the model's generalization. To address this issue, we propose using genetic programming (GP) to automatically construct new features with higher discriminant power than the original radiomic features. To achieve this goal, we introduce a fitness function that measures the classification performance ratio of output to input. The constructed features are then input for various classifiers to predict the two-year survival of NSCLC patients from two public CT datasets. Our approach is compared against two popular feature selection methods in radiomics to choose relevant radiomic features, and two GP-based feature construction methods whose fitness functions are based on measuring the constructed features' quality. The experimental results show that survival prediction models trained on GP-based constructed features outperform feature selection methods. Also, maximizing the classification performance gain output-to-input ratio produces features with higher discriminative power than only maximizing the classification accuracy from constructed features. Furthermore, a survival analysis demonstrated statistically significant differences between survival and non-survival groups in the Kaplan–Meier curves. Therefore, the proposed approach can be used as a complementary method for oncologists in determining the clinical management of NSCLC patients. [ABSTRACT FROM AUTHOR]

Purpose: Computer‐aided diagnosis (CAD) systems on breast ultrasound (BUS) aim to increase the efficiency and effectiveness of breast screening, helping specialists to detect and classify breast lesions. CAD system development requires a set of annotated images, including lesion segmentation, biopsy results to specify benign and malignant cases, and BI‐RADS categories to indicate the likelihood of malignancy. Besides, standardized partitions of training, validation, and test sets promote reproducibility and fair comparisons between different approaches. Thus, we present a publicly available BUS dataset whose novelty is the substantial increment of cases with the above‐mentioned annotations and the inclusion of standardized partitions to objectively assess and compare CAD systems. Acquisition and Validation Methods: The BUS dataset comprises 1875 anonymized images from 1064 female patients acquired via four ultrasound scanners during systematic studies at the National Institute of Cancer (Rio de Janeiro, Brazil). The dataset includes biopsy‐proven tumors divided into 722 benign and 342 malignant cases. Besides, a senior ultrasonographer performed a BI‐RADS assessment in categories 2 to 5. Additionally, the ultrasonographer manually outlined the breast lesions to obtain ground truth segmentations. Furthermore, 5‐ and 10‐fold cross‐validation partitions are provided to standardize the training and test sets to evaluate and reproduce CAD systems. Finally, to validate the utility of the BUS dataset, an evaluation framework is implemented to assess the performance of deep neural networks for segmenting and classifying breast lesions. Data Format and Usage Notes: The BUS dataset is publicly available for academic and research purposes through an open‐access repository under the name BUS‐BRA: A Breast Ultrasound Dataset for Assessing CAD Systems. BUS images and reference segmentations are saved in Portable Network Graphic (PNG) format files, and the dataset information is stored in separate Comma‐Separated Value (CSV) files. Potential Applications: The BUS‐BRA dataset can be used to develop and assess artificial intelligence‐based lesion detection and segmentation methods, and the classification of BUS images into pathological classes and BI‐RADS categories. Other potential applications include developing image processing methods like despeckle filtering and contrast enhancement methods to improve image quality and feature engineering for image description. [ABSTRACT FROM AUTHOR]

The computer vision community has proposed numerous formulations for object description based on human perceptivity and vast knowledge of the problem domain. In order to reduce human intervention, deep learning techniques are widely used to learn features automatically. However, they lack the property of explainability; that is, a human being understands the meaning of all the parts that make up the calculation of a feature. In this paper, we empirically show how Artificial Intelligence can automatically discover explainable formulations of a topological feature for binary images called the Euler characteristic. The training images are represented by bit-quad patterns, and a single-layer artificial neural network automatically learns the optimal combination of bit-quads to provide valid formulations to correctly calculate the Euler characteristic. We report the results on binary images of different complexities and sizes and compare them with state-of-the-art machine learning algorithms. Finally, we present 14 new equations to calculate the Euler number never reported in literature. [ABSTRACT FROM AUTHOR]

Audio files are frequent targets of malicious users who seek illegal profit trading with fake-quality content. For increasing the confidence in the integrity of audio files, the detection of fake-quality content is an important task. This paper proposes a method for detecting Advanced Audio Coding (AAC) compression on suspicious WAV files, in which the variance of the Modified Discrete Cosine Transform (MDCT) characterises four compression bitrates: uncompressed, 64 kbps, 128 kbps, and 256 kbps. This scheme takes advantage of the reduction of the variance of the high-frequency MDCT coefficients in compressed signals. Data obtained from MDCT coefficients generate a high-dimensional feature space. Hence, Principal Component Analysis, followed by Linear Discriminant Analysis, is used for projecting the high-dimensional data onto a lower-dimensional space. Besides, six supervised learning algorithms are compared for classifying four compression bitrates. The experiments show that using audio samples with 20 seconds and 1024 MDCT coefficients, an accuracy of 93% is reached with a Bayesian classifier. Collaterally, the detection between uncompressed and compressed signals attains an accuracy of 97% with Multinomial Logistic Regression. In conclusion, the proposed approach can detect previous AAC compression and can be potentially used when it is unfeasible to recover the suspicious signal completely. [ABSTRACT FROM AUTHOR]

So-called Vertex Chain Codes have been widely used to describe the shape of the objects. From these codes, several describing features can be obtained, e.g., the Euler characteristic. In this research, we show how Vertex Chain Codes can be used to train an Artificial Neural Network to compute the Euler characteristic of a 2-D binary image. We experimentally demonstrate how a simple linear neuron is enough to attain the goal. We present results with sets of 2-D binary images and objects of different complexity and size. [ABSTRACT FROM AUTHOR]