Deep learning model based on CNN-former in the diagnosis and detection of liver fibrosis

Background: Hepatic fibrosis is a common pathological process in the process of various liver injury and repair. It can develop into advanced cirrhosis or even liver cancer under the stimulation of various factors. Early diagnosis and intervention can significantly improve liver function and even reverse the histopathological process. Accurate non-invasive identification and staging of early liver fibrosis has become the research goal of many scholars. Objective: This study aimed to propose a model training method and system for liver fibrosis lesion detection based on CNN-Former. Study design: We first collect the clinical patient data and introduce the setup and data preprocessing of the experimental data, then introduce the experimental process of the feature engineering , the ablation experiment conducted and the evaluation index . Finally, compared this method with traditional method . Results: : By modeling the physiological and biochemical characteristics of patients with liver fibrosis, the pre-detection of liver fibrosis is carried out with high efficiency, high quality and high standard, providing a solution for existing research to rely on image information. In addition, on the basis of the existing CNN-form model, multi-CNN-form adds a multi-task mechanism to enable the model to predict liver fibrosis and related complications at the same time, providing a solution to the problem that traditional models focus on specific diseases and poor transplantation. Also, the accuracy, recall, AUC value, F1 value and other evaluation indexes of the model were calculated based on the test set and confusion matrix to compare the effects of various models in predicting liver fibrosis pre-detection. Finally, the effectiveness of this method compared with the traditional method is demonstrated from many angles. This makes it more suitable in practical application scenarios. Conclusions: : This paper mainly introduces a liver fibrosis pre-detection method based on deep learning, that is, the liver fibrosis pre-detection method based on CNN-former, which provides a solution for the existing research to rely on imaging information. In addition, the pre-detection method of liver fibrosis based on multi-CNN-former provides a solution to the problem of focusing on specific diseases and poor transplantation of traditional models.