NCT04299919

The AI Prognostic Assessment and Pathological Basis Research of Early HCC After Minimally Invasive Treatment

The Artificial Intelligent Prognostic Assessment and Pathological Basis Research of Early Primary Hepatocellular Carcinoma After Minimally Invasive Treatment Based on Multimodal MRI and Clinical Big Data

• Status: Unknown
• Phase: —
• Study type: Observational
• Enrollment: 1,200 (estimated)
• Sponsor: The First Affiliated Hospital of Dalian Medical University · Academic / Other
• Sex: All
• Age: —
• Healthy volunteers: Not accepted

Summary

The study evaluates artificial intelligence method based on multimodal magnetic resonance imaging (MRI) images and clinical data in preoperative prediction of prognosis in early hepatocellular carcinoma (HCC) patients treated with minimally invasive treatment. The correlation between prognosis-related MRI features and pathological features was studied through artificial intelligence method, so as to provide the interpretability of image features for predicting the prognosis of HCC patients treated with minimally invasive treatment.

Detailed description

The prognosis prediction of early stage hepatocellular carcinoma (HCC) after minimally invasive treatment involves clinical decision of treatment and follow-up. Magnetic resonance imaging (MRI) has become the main approach for monitoring and following up of HCC, however it's difficult to predict HCC prognosis before surgery. We found the following limitations among previous researches: multimodal MRI using different sequences shows uncertain boundaries of HCC, which makes precise segmentation more difficult, and also leads to an additional workload for extracting high throughput radiomics features, which are limited in quantity and repeatability. Regarding to prognosis aspect, the MRI images, clinical data, and follow up information have not been fully exploited yet. In addition, the prognosis result obtained by radiomics workflow is difficult to be explained and applied to clinical application. Therefore, we conduct a study to solve the problems mentioned above: (1) To explore an effective deep learning neural network method and a pre-training model for improving tumor segmentation accuracy. (2) To establish a method for extracting high-throughput multi-dimensional and multimodal MRI radiomics features related to HCC prognosis. (3) To explore a correlation between "multimodal MRI based pathological features of early stage HCC" and the results of "multimodal MRI based prognosis depth network of early stage HCC after minimally invasive treatment". Based on above approaches, we aim to establish "multimodal MRI based prognosis model of early stage HCC after minimally invasive treatment" in different clinical application scenarios guiding to clinical decision-making. Moreover, we also aim to explore the correlation between MRI radiomics features and pathology, which provides theoretical foundations for the MRI radiomics based pathological researches.

Conditions

• Hepatocellular Carcinoma

Interventions

Type	Name	Description
PROCEDURE	Minimally invasive treatment	All hepatocellular carcinoma (HCC) patients received minimally invasive treatment, including transcatheter arterial chemoembolization (TACE), radiofrequency ablation (RFA) or combined.
PROCEDURE	Hepatectomy	All hepatocellular carcinoma (HCC) patients received hepatectomy.

Timeline

• Start date: 2007-04-01
• Primary completion: 2023-06-30
• Completion: 2024-06-30
• First posted: 2020-03-09
• Last updated: 2020-03-09

Locations

1 site across 1 country: China