Effect of Tissue Excitation in Breast Cancer Detection from Ultrasound RF Time Series: Phantom studies

Breast cancer is considered as the most common type of cancer among women. The treatment can be more successful when the lesion and its type are diagnosed in the early stages. A biopsy, which is invasive and causes a lot of psychological and financial costs on the patient, can be used to ensure the type of lesions due to the lack of sufficient accuracy of each the previously used method. Thus, a low-cost, non-radioactive, and non-invasive method should be implemented for determining the type of lesion. Based on the recent studies on RF time series, the interactions of this signal with tissue and echo patterns can create useful information about the propagation environment. In this study, RF signals were recorded while stimulating breast tissue, and afterwards the effect of this interaction on the classification of cancerous and non-cancerous lesions was observed by extracting appropriate features from the RF time series. A Supersonic Imagine US system equipped with a linear probe was applied for recording the signal. The data were recorded from two types of agar-gelatin phantom mimicking breast tissue. The data were collected under a fixed probe, as well as static and vibrational excitation. Finally, SVM classifiers were used to classify cancerous and healthy tissues. In addition, the effect of central frequency on this classification was considered. Based on the results, tissue characteristics can add more information to the RF time series and improve classification results. The best result was obtained by 6.4 central frequency ultrafast beamformed data with 98.78±2.77 % accuracy.