PaperlayerView exact revision on arXiv

Source-linked AI summary

On the Role of MRI Sequences in Cross-Dataset Generalization for Brain Tumor Segmentation

Henrique Zan Grande, João G. Pitol, Lucas B. Schuck, Rafael V. Serenato, Rayson Laroca, Andre Gustavo Hochuli

arXiv:2608.29944v1cs.CV

TL;DR

Cross-dataset brain tumor segmentation remains difficult under domain shift and limited annotations. The study independently evaluates MRI modalities with a controlled ResUNet-based protocol, then tests limited target-domain fine-tuning. T2f/FLAIR provides the strongest cross-dataset performance, while 10–20% target data already approaches upper-bound performance.

  • Problem

    Cross-dataset brain tumor segmentation is challenged by domain shift and limited annotated medical data, motivating study of modality-specific robustness and annotation-efficient adaptation.

    Source: §I · p. 1, §III · p. 2

  • Method

    The study uses a standardized 2D ResUNet pipeline to train each MRI modality independently, evaluate controlled transfer settings, and fine-tune the best modality with limited target-domain data.

    Source: §I · p. 1, B. Architecture · p. 3

  • Results

    T2f/FLAIR achieves the strongest cross-dataset performance, while fine-tuning with 10–20% of target-domain data already approaches upper-bound performance.

    Source: A. Generalization Evaluation (RQ1/RQ2) · pp. 3–4, B. Domain Adaptation (RQ3) · p. 4

  • Takeaways & Limitations

    Modality choice, data diversity, and lesion characteristics influence generalization, while limited target-domain adaptation can substantially improve performance with reduced annotation and computational costs.

    Source: §V · pp. 5–6

Abstract

from arXiv · show

Brain tumor segmentation in magnetic resonance imaging (MRI) is a critical task for diagnosis and treatment planning. Despite the success of deep learning architectures such as U-Net and its variants, performance degradation across datasets remains a major challenge, particularly under domain shift and limited annotated data. To address this issue, this study systematically evaluates how individual MRI sequences influence model robustness across two well-known datasets. A ResUNet-based framework is employed, where each modality is trained independently to isolate its effect under a controlled cross-dataset evaluation protocol with tumor size stratification, without target-domain training, or with limited domain adaptation. Results show that the T2f/FLAIR sequence achieves the best cross-dataset performance, with Dice scores exceeding 75%. It consistently outperforms other modalities across most tumor size ranges, while multi-sequence training further improves performance. Additionally, even limited target-domain adaptation yields rapid initial gains, reducing the need for extensive annotations and costly retraining. Our source code is publicly available at https://github.com/henrique-zan/brain_tumor_segmentation/.

I. INTRODUCTION

Brain tumor segmentation is clinically important, but domain shift and limited annotations can degrade cross-dataset performance. This study isolates MRI-sequence effects and evaluates modality robustness, generalization gaps, and data-efficient adaptation using a controlled ResUNet pipeline.

  • Brain tumor segmentation supports diagnosis, surgical planning, and clinical follow-up in medical imaging.Source: §I · p. 1
  • Domain shifts from acquisition protocols, scanners, patient populations, and image characteristics can substantially degrade performance on unseen datasets.Source: §I · p. 1
  • Fewer studies have isolated each MRI sequence under cross-dataset domain shift, despite the potential to identify more transferable representations.Source: §I · p. 1
  • The study asks how modality affects cross-domain degradation, which sequences are most robust, and how target-domain annotation effort relates to improvement.Source: §I · p. 1
  • Each modality is trained and evaluated independently with a standardized 2D ResUNet pipeline, followed by fine-tuning of the best-performing modality using limited target-domain data.Source: §I · p. 1
  • The paper contributes controlled modality-specific robustness evaluation, quantitative generalization-gap analysis, robust-sequence identification, and annotation-budget assessment.Source: §I · p. 1

II. LITERATURE REVIEW

Brain tumor segmentation research is dominated by U-Net-based convolutional methods and multimodal fusion, but cross-dataset robustness remains insufficiently addressed. This study targets the unquantified modality-specific contribution under domain shift and limited annotation budgets.

  • U-Net and its variants dominate MRI brain tumor segmentation, while prior work emphasizes architectural improvements and multimodal fusion.Source: §II · p. 2
  • U-Net combines contextual representation with spatial detail through its encoder-decoder structure and skip connections.Source: §II · p. 2
  • ResUNet variants improve optimization or feature extraction through residual blocks, attention mechanisms, and automated configuration.Source: §II · p. 2
  • Combining T1, T1c, T2, and T2f/FLAIR can improve in-domain segmentation accuracy and robustness over single-sequence inputs.Source: §II · p. 2
  • Individual MRI-sequence contributions to cross-domain robustness remain insufficiently quantified, especially under limited annotation budgets.Source: §II · p. 2

III. PROBLEM STATEMENT

The problem is to measure segmentation loss when a single-sequence model transfers between BraTS cohorts, identify more robust sequences, and assess how small labeled target subsets compensate for domain shift. A controlled evaluation uses identical modeling and preprocessing conditions across modalities.

  • Cross-dataset transfer can lose segmentation performance, motivating comparisons of modality robustness and annotation-efficient compensation.Source: §III · p. 2
  • Every modality uses the same 2D ResUNet architecture, data preparation, and optimization settings for controlled comparison.Source: §III · p. 2
  • The comparison includes in-domain evaluation, direct cross-dataset transfer, joint multi-dataset training, and incremental target-domain fine-tuning.Source: §III · p. 2
  • The experimental design applies uniform partitioning, preprocessing, architecture, training configuration, and evaluation procedures.Source: §III · p. 2

A. Datasets

BraTS 2020 and BraTS 2024 provide complementary cohorts for testing robustness under realistic dataset differences. Patient-level partitioning, uniform resizing, tumor-containing slice selection, and conservative augmentation standardize preparation.

  • BraTS 2020 contains 369 multimodal MRI studies, whereas BraTS 2024 includes approximately 1,350 cases in a larger, more heterogeneous cohort.Source: A. Datasets · p. 2
  • Figure 1 presents representative BraTS 2020 cases above BraTS 2024 cases across four MRI sequences and their binary masks.Source: Figure 1 · p. 2
  • The datasets differ in patient characteristics, tumor morphology, scanners, and acquisition protocols, creating realistic domain shift.Source: A. Datasets · p. 2
  • Patient-level splits assign 70% of patients to training, 15% to validation, and 15% to testing, preventing slice-level leakage across subsets.Source: A. Datasets · p. 3
  • Images are resized to 240 × 240 pixels, and 2D slices containing tumor tissue are extracted for each modality.Source: A. Datasets · p. 3
  • Training augmentation uses 90◦ rotations and horizontal and vertical flips, excluding aggressive intensity and geometric transformations.Source: A. Datasets · p. 3

B. Architecture

The study uses a 2D ResUNet with a pretrained ResNet-34 encoder, residual learning, skip connections, and a decoder that produces binary segmentation maps from single-channel MRI slices.

Architecture

  • The 2D ResUNet uses a pretrained ResNet-34 encoder with residual mappings and skip connections for hierarchical, multi-scale feature extraction.Residual identity mappings support gradient propagation, while skip connections preserve fine-grained spatial information for decoding.Source: B. Architecture · p. 3
  • The decoder has four upsampling stages that progressively refine concatenated encoder features through transposed convolutions and convolutional blocks.Its channel progression is 512 →256 →128 →64 →64.Source: B. Architecture · p. 3
  • The network receives a single-channel MRI slice and outputs a binary segmentation map through a final 1 × 1 convolution.Dropout with p = 0.2 is used within convolutional blocks to reduce overfitting in limited-data and cross-dataset settings.Source: B. Architecture · p. 3
  • Each modality is trained independently with a combined Dice-loss and binary-crossentropy objective under early stopping and adaptive learning-rate reduction.Training used up to 80 epochs, batch size 32, and an initial learning rate of 1×10−3.Source: B. Architecture · p. 3

IV. EXPERIMENTS AND DISCUSSION

Experiments compare segmentation across BraTS 2020 and BraTS 2024 using multiple training and evaluation protocols, overlap metrics, target-domain fine-tuning, and tumor-size stratification.

Experimental Design

  • The experiments use BraTS 2020 and BraTS 2024 to compare in-domain, cross-dataset, and multi-domain evaluation protocols.Fine-tuning with increasing fractions of labeled target-domain data addresses the domain-adaptation question.Source: §IV · p. 3

Reproducibility

  • The source code, trained weights, models, and experimental artifacts are publicly available for reproducibility.The repository is linked at github.com/henrique-zan/brain_tumor_segmentation/.Source: §IV · p. 3

Tumor-Size Analysis

  • Tumor-size quartiles and representative cases are analyzed because Dice is more sensitive to small absolute errors in small lesions.This stratification examines whether modality performance varies with tumor size.Source: §IV · p. 3

Evaluation Metrics

  • All comparisons report Dice and Intersection over Union, averaged over tumor-containing test slices.Both metrics quantify agreement between predicted and reference tumor masks.Source: §IV · p. 3

A. Generalization Evaluation (RQ1/RQ2)

The generalization evaluation progressively increases distributional discrepancy across in-domain, direct cross-dataset, and joint multi-domain protocols. T2F/FLAIR shows the strongest cross-domain robustness, while multi-domain training reduces degradation.

Protocol Design

  • The evaluation protocols progressively increase distributional discrepancy to assess modality-specific generalization behavior.This design supports controlled comparison of robustness under changing train-test distributions.Source: A. Generalization Evaluation (RQ1/RQ2) · p. 3

Baseline

Cross-Dataset

  • Direct cross-dataset evaluation trains on one dataset and tests on the other without adaptation, using both transfer directions.The protocol introduces differences in acquisition protocols, intensity distributions, and population characteristics.Source: A. Generalization Evaluation (RQ1/RQ2) · p. 3

Multi-Domain Training

  • Joint training uses balanced 50/50 data from BraTS 2020 and BraTS 2024 to test whether heterogeneous exposure improves generalization.Patient-level holdout splits are preserved to avoid data leakage.Source: A. Generalization Evaluation (RQ1/RQ2) · p. 3

Results

  • T2F maintains the highest cross-domain performance, up to 76.2 Dice, while T1-based modalities show lower robustness at approximately 49.7 Dice.In-domain Dice values were 77.3–78.3 for FLAIR and 73.3–74.2 for T2W, but cross-domain transfer caused substantial degradation.Source: A. Generalization Evaluation (RQ1/RQ2) · pp. 3–4
  • Multi-domain training further mitigates degradation, with T2F reaching 75.0 compared with 77.3/78.3 in the corresponding single-domain baselines.The results support more transferable representations for T2-based inputs than for T1-based inputs.Source: A. Generalization Evaluation (RQ1/RQ2) · p. 4

B. Domain Adaptation (RQ3)

Fine-tuning the best-performing T2F/FLAIR modality with limited labeled target-domain data produces rapid gains in cross-domain performance, followed by diminishing returns.

Domain Adaptation

  • Fine-tuning with only 10–20% of labeled target-domain data already approaches upper-bound performance.Additional target data produces a plateau, indicating diminishing returns.Source: B. Domain Adaptation (RQ3) · p. 4
  • Limited target-domain adaptation substantially improves generalization while reducing annotation and computational costs.The largest gains occur when adapting BraTS 2020-trained models to BraTS 2024; the reverse direction yields more modest improvements of approximately 5% Dice.Source: B. Domain Adaptation (RQ3) · p. 4

C. Qualitative Analysis

Tumor-size-stratified results show that overlap metrics improve from smaller to larger lesions, with T2f/FLAIR strongest for small tumors and T2w slightly favored for the largest.

Tumor-size stratification

  • Dice and IoU increase consistently from the smallest tumor quartile (Q1) to the largest (Q4).The quartiles contain equally populated groups, and values in parentheses report mean tumor area.Source: C. Qualitative Analysis · p. 4, Table II · p. 5

Modality comparison

  • T2f/FLAIR performs best for small tumors, indicating higher sensitivity to subtle boundaries.For medium tumors, T2f/FLAIR and T2w are comparable and both outperform T1-based modalities.Source: C. Qualitative Analysis · p. 5
  • T2w slightly surpasses T2f/FLAIR for large tumors in the upper quartile.For larger volumes overall, the two modalities offer similar performance, with a slight T2w advantage in Q4.Source: C. Qualitative Analysis · p. 5

Qualitative examples

  • Figure 5 presents representative segmentation masks for cases closest to the mean tumor size in each quartile.These masks complement the quantitative tumor-size analysis in Table II.Source: C. Qualitative Analysis · p. 5, Figure 5 · p. 5
  • Figure 6 contrasts high-accuracy, intermediate, and failure cases using T2f/FLAIR predictions against ground-truth overlays.The reported Dice scores are 99.1%, 70.1%, and 0.0%, respectively.Source: Figure 6 · p. 5, C. Qualitative Analysis · p. 5

V. CONCLUSIONS

The study concludes that MRI modality, training-domain diversity, and tumor size shape cross-dataset generalization. T2f/FLAIR is strongest overall, while limited target-domain fine-tuning reaches near-upper-bound performance efficiently.

Conclusions

  • T2f/FLAIR provides the best cross-dataset performance and strongest generalization, followed closely by T2w.T1-based modalities exhibit lower robustness under domain shift.Source: §V · p. 5
  • Multi-domain training reduces the generalization gap and brings transfer performance closer to in-domain conditions.The conclusion identifies data diversity as an important factor in cross-dataset performance.Source: §V · p. 5
  • Tumor size significantly affects performance: T2f/FLAIR is more effective for small lesions, while T2w becomes comparable or slightly superior for larger tumors.The modality-size relationship is reported as part of the study’s conclusions.Source: §V · pp. 5–6
  • 10–20% of target-domain data already approaches upper-bound performance during fine-tuning, with limited gains thereafter.This indicates diminishing returns as the annotation fraction increases.Source: §V · p. 6
  • The findings provide practical guidance for modality selection and resource-constrained deployment scenarios.The paper links generalization to modality choice, data diversity, and lesion characteristics.Source: §V · p. 6

Future work

  • Future work includes multimodal and 3D models, independent clinical datasets beyond BraTS, improved small-lesion detection, and foundation-model evaluation.The proposed directions also include improved embedding spaces and adaptive architectures.Source: §V · p. 6

Related papers

Loading 2608.29944v1…