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

TL;DR

MTI-Net addresses the absence of an end-to-end framework for jointly analyzing liver tumors across structural, functional, and diagnostic dimensions. It introduces MdIEF for entropy-guided fusion of spatial and spectral high-frequency information, TIM for cross-task consistency between segmentation and regression, and TDD for adversarial alignment of regression and classification via a Transformer-based discriminator. A shallow Transformer handles sequence-level relations across four dynamic MRI phases. Evaluated on 238-subject data, MTI-Net achieves state-of-the-art performance across segmentation, regression, and classification, illustrating strong inter-task synergy and potential clinical impact.

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.

Figures (4)

• Figure 1: From left to right, (a) shows the difference of dynamic enhancement in the time-intensity curve between hemangioma and HCC. (b) and (c) show the advantages of our method compared to the clinical method.
• Figure 2: Overview of the proposed MTI-Net. It provides a novel framework to achieve multi-task mutual promotion via the proposed MdIEF, TIM and TDD.
• Figure 3: The visual example of liver tumor segmentation, where the last row of the comparison and ablation study shows our MTI-Net is the best.
• Figure 4: Confusion matrix of comparison for evaluating classification.