Abstract
Brain metastasis segmentation from multiparametric magnetic resonance imaging (MRI) is challenging because lesions vary in size, appearance, multiplicity, and treatment status. We present the Gated Frequency-Enhanced Class-Wise Ensemble (GFCE) for the Brain Tumor Segmentation (BraTS) 2026 Challenge. GFCE integrates two complementary nnUNetv2 residual encoder-decoder streams: a baseline model and a frequency-enhanced branch that applies a lightweight gated adaptation of Adaptive Laplacian Convolution (ALC) to the multimodal input. Their whole-tumor (WT), tumor-core (TC), enhancing-tumor (ET), and resection-cavity (RC) probability maps are fused with class-specific weights and refined by region-specific thresholding and connected-component filtering. On the official validation cohort, GFCE achieved lesion-wise Dice scores of 0.709, 0.685, 0.652, and 0.586 for enhancing tumor, tumor core, whole tumor, and resection cavity, respectively. Instance-level detection F1 was 0.806, 0.768, 0.773, and 0.314 for the same regions. These results demonstrate the efficacy of our proposed class-wise ensembling strategy while also underscoring the ongoing difficulty in accurately segmenting small lesions and resection cavities.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/BraTS_METS_011.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/profile?id=%7ERuifeng_Wang5
BibTex
@InProceedings{WanRui_Gated_MICCAISAT2026,
author = { Wang, Ruifeng AND Peng, Yanjun AND Chen, Xue},
title = { { Gated Frequency-Enhanced Class-Wise Ensemble for Brain Metastases Segmentation } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17253},
month = {pending},
page = {pending}
}