Prediction of programmed cell death protein 1 in hepatocellular carcinoma patients using radiomics analysis with radiofrequency-based ultrasound multifeature maps

Abstract

Background This study explored the feasibility of radiofrequency (RF)-based radiomics analysis techniques for the preoperative prediction of programmed cell death protein 1 (PD-1) in patients with hepatocellular carcinoma (HCC). Methods The RF-based radiomics analysis method used ultrasound multifeature maps calculated from the RF signals of HCC patients, including direct energy attenuation (DEA) feature map, skewness of spectrum difference (SSD) feature map, and noncentrality parameter S of the Rician distribution (NRD) feature map. From each of the above ultrasound maps, 345 high-throughput radiomics features were extracted. Then, the useful radiomics features were selected by the sparse representation method and input into support vector machine (SVM) classifier for PD-1 prediction. Results and conclusion Among all the RF-based prediction models and the ultrasound grayscale comparative model, the RF-based model using all of the three ultrasound feature maps had the highest prediction accuracy (ACC) and area under the curve (AUC), which were 92.5% and 94.23%, respectively. The method proposed in this paper is effective for the meaningful feature extraction of RF signals and can effectively predict PD-1 in patients with HCC.

Rapid intraoperative multi-molecular diagnosis of glioma with ultrasound radio frequency signals and deep learning

“…Two classic baseline models were chosen as references based on the study approach of ultrasound RF signals. The first model is the combination of RF multi-feature maps and radiomics proposed by Wang Q. et al, 27 which we refer to as Maps-Radiomic. The model requires an artificially defined functions and a tumor mask to extract signal features within the tumor.…”

“…In recent years, clinical diagnostic studies based on RF signals have emerged, and they mainly focus on breast, 23 , 24 liver, 25 , 26 , 27 , 28 thyroid, 29 , 30 and prostate. 31 These studies can be categorized into two main approaches.…”

“… 31 These studies can be categorized into two main approaches. The first approach 23 , 24 , 25 , 26 , 27 is to perform complex pre-processing of the RF signal, extracting specific components and utilizing radiomics or deep learning model for post-processing. The second category 28 , 29 , 30 , 31 is the direct analysis of the normalized RF signal using deep learning models.…”

“…The two categories have achieved promising results in the classification of benign and malignant tumors, molecular diagnosis, and prognostic prediction. Previous studies 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 have demonstrated that subtle changes in histological structure reflected in ultrasound signals can unveil the presence of tumor and mutations in molecular biomarkers. The ultrasound RF signal, rich in information, theoretically offers advantages over traditional ultrasound images in distinguishing molecular subtypes.…”