The Value of Convolutional Neural Network-Based Magnetic Resonance Imaging Image Segmentation Algorithm to Guide Targeted Controlled Release of Doxorubicin Nanopreparation.

Hui Gao,Hujun Liu,Fei Jia

doi:10.1155/2021/9032017

Abstract

There was an investigation of the auxiliary role of convolutional neural network- (CNN-) based magnetic resonance imaging (MRI) image segmentation algorithm in MRI image-guided targeted drug therapy of doxorubicin nanomaterials so that the value of drug-controlled release in liver cancer patients was evaluated. In this study, 80 patients with liver cancer were selected as the research objects. It was hoped that the CNN-based MRI image segmentation algorithm could be applied to the guided analysis of MRI images of the targeted controlled release of doxorubicin nanopreparation to analyze the imaging analysis effect of this algorithm on the targeted treatment of liver cancer with doxorubicin nanopreparation. The results of this study showed that the upgraded three-dimensional (3D) CNN-based MRI image segmentation had a better effect compared with the traditional CNN-based MRI image segmentation, with significant improvement in indicators such as accuracy, precision, sensitivity, and specificity, and the differences were all statistically marked (p < 0.05). In the monitoring of the targeted drug therapy of doxorubicin nanopreparation for liver cancer patients, it was found that the MRI images of liver cancer patients processed by 3D CNN-based MRI image segmentation neural algorithm could be observed more intuitively and guided to accurately reach the target of liver cancer. The accuracy of targeted release determination of nanopreparation reached 80 ± 6.25%, which was higher markedly than that of the control group (66.6 ± 5.32%) (p < 0.05). In a word, the MRI image segmentation algorithm based on CNN had good application potential in guiding patients with liver cancer for targeted therapy with doxorubicin nanopreparation, which was worth promoting in the adjuvant treatment of targeted drugs for cancer.

Highlights

Liver cancer refers to cancer that occurs in the liver and is one of the most common tumors in China [1, 2]
All patients were rolled randomly into two groups and were examined with magnetic resonance imaging (MRI) scans. e experimental group received a convolutional neural network- (CNN-)based MRI image segmentation algorithm for MRI image processing and analysis, while the routine artificial MRI image analysis method was applied to the control group. is study was approved by the Ethics Committee of the hospital, and the patients and their family members included in the study were all informed and signed informed consent forms
Due to the shortcomings in the fitting effect of a single algorithm, this research aimed to integrate the ensemble learning (EL) algorithm on this basis to systematically reduce the generalization error of the model through the integration of multiple models. e Extreme Gradient Boosting (XGBoost) algorithm in the EL algorithm was brought into the model. e classification and regression trees (CART) were adopted by the XGBoost algorithm as the weak classification item

Summary

Introduction

Liver cancer refers to cancer that occurs in the liver and is one of the most common tumors in China [1, 2]. As for the disease distribution data, the incidence of liver cancer in rural areas is higher than that in urban areas, and the incidence of liver cancer in males is higher than that in females [3,4,5]. Diagnostic methods for liver cancer include imaging examinations (dynamic contrast-enhanced magnetic resonance imaging (MRI), dynamic contrast-enhanced computed tomography (CT), and selective hepatic arteriography), the alpha-fetoprotein (AFP) indicators in blood-drawn results, and liver lesion tissue biopsy. The advantage of MRI imaging in the diagnosis of liver cancer lies in its high sensitivity, which can display the tissue structure of liver cancer more clearly than CT examination. Ere are differences in treatment methods for different stages and different types of liver cancer. MRI examination adds coronal and sagittal examination based on the cross section, which can reduce the interference of artifacts to a certain extent [7,8,9]. ere are differences in treatment methods for different stages and different types of liver cancer. e currently available treatment methods include

Objectives

Methods

Results

Discussion

Conclusion