Applications of machine-learning algorithms for prediction of benign and malignant breast lesions using ultrasound radiomics signatures: A multi-center study‏

The present study examines the diagnostic efficacy of machine-learning algorithms in differentiating benign and malignant breast lesions using ultrasound images...

Abstract

Artificial intelligence (AI) algorithms have an enormous potential to impact the field of radiology and diagnostic imaging, especially the field of cancer imaging. There have been efforts to use AI models to differentiate between benign and malignant breast lesions. However, most studies have been single-center studies without external validation. The present study examines the diagnostic efficacy of machine-learning algorithms in differentiating benign and malignant breast lesions using ultrasound images. Ultrasound images of 1259 solid non-cystic lesions from 3 different centers in 3 countries (Malaysia, Turkey, and Iran) were used for the machine-learning study. A total of 242 radiomics features were extracted from each breast lesion, and the robust features were considered for models’ development. Three machine-learning algorithms were used to carry out the classification task, namely, gradient boosting (XGBoost), random forest, and support vector machine. Sensitivity, specificity, accuracy, and area under the ROC curve (AUC) were determined to evaluate the models. Thirty-three robust features differed significantly between the two groups from all of the features. XGBoost, based on these robust features, showed the most favorable profile for all cohorts, as it achieved a sensitivity of 90.3%, specificity of 86.7%, the accuracy of 88.4%, and AUC of 0.890. The present study results show that incorporating selected robust radiomics features into well-curated machine-learning algorithms can generate high sensitivity, specificity, and accuracy in differentiating benign and malignant breast lesions. Furthermore, our results show that this optimal performance is preserved even in external validation datasets.

The flow diagram of the multistep process of image segmentation, feature extraction, model development, and performance evaluation utilized in the present study.
Picture of Ardakani AA

Ardakani AA

He received his Ph.D. in Medical Physics in 2018 from the Iran University of Medical Sciences (IUMS), specializing in medical imaging and using artificial intelligence in radiological diagnosis. His research interests focus on the physics of medical imaging systems, quantitative analysis of medical images, and applying artificial intelligence in diagnostic radiology procedures. He is an assistant professor of Medical Physics at Shahid Beheshti University of Medical Sciences.

Leave a Reply

Your email address will not be published. Required fields are marked *