A Transformer-Based microvascular invasion classifier enhances prognostic stratification in HCC following radiofrequency ablation

• Published in: Liver international 2024-04, Vol.44 (4), p.894-906
• Main Authors: Wang, Wentao, Wang, Yueyue, Song, Danjun, Zhou, Yingting, Luo, Rongkui, Ying, Siqi, Yang, Li, Sun, Wei, Cai, Jiabin, Wang, Xi, Bao, Zhen, Zheng, Jiaping, Zeng, Mengsu, Gao, Qiang, Wang, Xiaoying, Zhou, Jian, Wang, Manning, Shao, Guoliang, Rao, Sheng-Xiang, Zhu, Kai
• Format: Article
• Language: English
• Subjects: Ablation; Biomarkers; Carcinoma, Hepatocellular; Cirrhosis; Classifiers; Deep learning; Hepatitis B; Hepatocellular carcinoma; Humans; Liver cancer; Liver Neoplasms - pathology; Magnetic resonance imaging; Medical imaging; Medical prognosis; Microvasculature; Neoplasm Invasiveness; Prognosis; Radio frequency; Radiofrequency Ablation; Retrospective Studies; Risk factors; Statistical analysis; Survival; Transformers
• ISSN: 1478-3223; 1478-3231; 1478-3231
• DOI: 10.1111/liv.15846

We aimed to develop a Transformer-based deep learning (DL) network for prognostic stratification in hepatocellular carcinoma (HCC) patients undergoing RFA. A Swin Transformer DL network was trained to establish associations between magnetic resonance imaging (MRI) datasets and the ground truth of microvascular invasion (MVI) based on 696 surgical resection (SR) patients with solitary HCC ≤3 cm, and was validated in an external cohort (n = 180). The multiphase MRI-based DL risk outputs using an optimal threshold of .5 was employed as a MVI classifier for prognosis stratification in the RFA cohort (n = 180). Over 90% of all enrolled patients exhibited hepatitis B virus infection. Liver cirrhosis was significantly more prevalent in the RFA cohort compared to the SR cohort (72.2% vs. 44.1%, p