Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma

Hepatocellular Carcinoma

magnetic resonance image features, pathological differentiation, machine learning-based model

Microvascular invasion (MVI) has been well demonstrated as an unfavorable prognostic factor for hepatocellular carcinoma (HCC), and patients with MVI have a high risk of tumor recurrence after curative hepatectomy. Currently, the diagnosis of MVI is determined on the postoperative histologic examination, which greatly limits its influence on preoperative decision making. Therefore, we constructed this prospective study to develop a machine learning-based model for preoperative prediction of MVI by extracting high-dimensional magnetic resonance (MR) image features.

Histologically-diagnosed primary HCC after curative hepatectomy. The magnetic resonance image will be imported into the imaging management software (GE healthcare Analysis-Kit software),and the tumor lesions will manually delineated by two independent radiologists and then reconstruct into three-dimensional images for feature extraction. The radiomic textural features including grayscale histogram, transform matrix, wavelet transform and filter transformation are automatically extracted by the Analysis-Kit software.The high-throughput extracted features will be then selected by the univariate analysis, and a prediction model will be developed based on machine learning algorithm in a training set in which patients were collected from a retrospective study. And in the present study, an independent validation set will be collected and used to validate the prediction accuracy of the model.

No related papers found yet.

China