biostudies-literatureWang LScience Health Joint Medical Scientific Research Project of ChongqingTsinghua UniversityNatural Science Foundation of Fujian ProvinceNational Natural Science Foundation of ChinaChinese Academy of Medical Sciences Initiative for Innovative Medicine106060https://www.ebi.ac.uk/biostudies/studies/S-EPMC12719682biostudies-literatureUnknown122<h4>Background</h4>Pathological evaluation of hepatocellular carcinoma (HCC) traditionally relies on surgical resection, posing risks of infection and complications while failing to provide comprehensive pathological insights preoperatively. This study aims to develop HepaPathGPT, which utilises preoperative imaging to deliver detailed pathological interpretations, enabling non-invasive, real-time pathological assessments for patients with HCC.<h4>Methods</h4>A retrospective study of 1091 patients with HCC from 10 independent cohorts was used. SegFormer-b5 segmented tumour regions, and vision-language alignment mapped imaging features to pathology descriptions. We fine-tuned four pretrained frameworks using Low-Rank Adaptation (LoRA) to efficiently translate imaging features into structureEBioMedicineA generative vision-language model for holistic pathological assessment using preoperative imaging in hepatocellular carcinoma.PMC12719682123266182022ZLA0072023MSXM0922023J060562019-I2M-5-05682473201820900538209005282272703Wu MTian FLi CHu ZFeng XLiao XCao JLi FHu XCai JLiu RWang JWang LHou LXia HDong JJin MWang XDu KZhao JYang SfalseA generative vision-language model for holistic pathological assessment using preoperative imaging in hepatocellular carcinoma.<h4>Background</h4>Pathological evaluation of hepatocellular carcinoma (HCC) traditionally relies on surgical resection, posing risks of infection and complications while failing to provide comprehensive pathological insights preoperatively. This study aims to develop HepaPathGPT, which utilises preoperative imaging to deliver detailed pathological interpretations, enabling non-invasive, real-time pathological assessments for patients with HCC.<h4>Methods</h4>A retrospective study of 1091 patients with HCC from 10 independent cohorts was used. SegFormer-b5 segmented tumour regions, and vision-language alignment mapped imaging features to pathology descriptions. We fine-tuned four pretrained frameworks using Low-Rank Adaptation (LoRA) to efficiently translate imaging features into structure2025-01-01T00:00:00Z2025 Dec2026-06-06T05:00:16.024Z2026-05-26T03:12:18.204ZS-EPMC127196824133793510.1016/j.ebiom.2025.106060