Generative AI-Driven Liver Reconstruction for Healthcare Applications in Consumer Electronics with Diffusion Model and Graph Neural Network

Magnetic hyperthermia therapy (MHT) is an emerging noninvasive treatment for liver cancer that depends on accurate digital liver reconstruction. However, the limited availability of annotated medical data, particularly for tumors and vascular structures, hinders effective model training. This challe...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on consumer electronics p. 1
Main Authors: Xu, Xu, Yang, Jing, Liu, Xiaoli, Khan, Muhammad Attique, Jiang, Weiwei, Baili, Jamel, Yee, Por Lip, Li, Congsheng
Format: Journal Article
Language:English
Published: IEEE 2025
Subjects:
Accuracy
Biomedical imaging
Consumer electronics
Consumer Electronics Healthcare
Deep Learning
Diffusion Model
Graph neural networks
Image reconstruction
Image segmentation
Liver
Magnetic Hyperthermia Therapy
Reconstruction
Segmentation
Surface reconstruction
Three-dimensional displays
Tumors
ISSN:0098-3063, 1558-4127
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Magnetic hyperthermia therapy (MHT) is an emerging noninvasive treatment for liver cancer that depends on accurate digital liver reconstruction. However, the limited availability of annotated medical data, particularly for tumors and vascular structures, hinders effective model training. This challenge is compounded by the need for individualized anatomical modeling in portable healthcare systems powered by consumer electronics. To overcome these challenges, we present a generative AI framework that integrates a conditional diffusion model with a graph neural network (GNN) to achieve high-fidelity, patient-specific liver reconstruction. Our framework combines a conditional diffusion model, which synthesizes realistic CT images to enrich liver anatomy, with a graph neural network that refines 3D surface reconstructions. Evaluated on public and clinical datasets, the method achieves higher segmentation accuracy and surface quality than existing approaches. By enhancing preoperative temperature-field simulations, the proposed approach supports individualized MHT planning and shows the potential of embedding generative AI in consumer healthcare devices.
ISSN:0098-3063
1558-4127
DOI:10.1109/TCE.2025.3631634