Adversarial Multi-Task Learning for Liver Tumor Segmentation, Dynamic Enhancement Regression, and Classification

2025-11-25

AI Summary

Existing methods struggle to jointly perform liver tumor segmentation, dynamic contrast-enhanced (DCE) MRI regression, and lesion classification within an end-to-end framework, primarily due to insufficient modeling of high-order inter-task dependencies and inadequate exploitation of temporal dynamics in DCE-MRI sequences. To address this, we propose MTI-Net, a unified multi-task learning architecture featuring: (1) a multi-domain information entropy fusion module that jointly encodes frequency- and spectrum-domain features; (2) a task-driven discriminator and a segmentation–regression high-order consistency constraint to explicitly model cross-task dependencies; and (3) a lightweight Transformer encoder for temporal relationship modeling in DCE-MRI, integrated with adversarial learning and multi-task optimization. Evaluated on 238 clinical cases, MTI-Net significantly outperforms both single-task and state-of-the-art multi-task baselines across all three tasks, thereby enhancing the accuracy and clinical reliability of liver tumor assessment.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Liver tumor segmentation, dynamic enhancement regression, and classification are critical for clinical assessment and diagnosis. However, no prior work has attempted to achieve these tasks simultaneously in an end-to-end framework, primarily due to the lack of an effective framework that captures inter-task relevance for mutual improvement and the absence of a mechanism to extract dynamic MRI information effectively. To address these challenges, we propose the Multi-Task Interaction adversarial learning Network (MTI-Net), a novel integrated framework designed to tackle these tasks simultaneously. MTI-Net incorporates Multi-domain Information Entropy Fusion (MdIEF), which utilizes entropy-aware, high-frequency spectral information to effectively integrate features from both frequency and spectral domains, enhancing the extraction and utilization of dynamic MRI data. The network also introduces a task interaction module that establishes higher-order consistency between segmentation and regression, thus fostering inter-task synergy and improving overall performance. Additionally, we designed a novel task-driven discriminator (TDD) to capture internal high-order relationships between tasks. For dynamic MRI information extraction, we employ a shallow Transformer network to perform positional encoding, which captures the relationships within dynamic MRI sequences. In experiments on a dataset of 238 subjects, MTI-Net demonstrates high performance across multiple tasks, indicating its strong potential for assisting in the clinical assessment of liver tumors. The code is available at: https://github.com/xiaojiao929/MTI-Net.

Problem

• Simultaneously performing liver tumor segmentation, enhancement regression, and classification
• Lack of effective framework capturing inter-task relevance for mutual improvement
• Absence of mechanism for effective dynamic MRI information extraction

Innovation

• Adversarial multi-task learning integrates segmentation, regression, classification
• Multi-domain entropy fusion uses frequency and spectral features
• Shallow Transformer captures dynamic MRI sequence relationships