BATS: Resource-Efficient Volumetric Segmentation with Boundary-Aware Mixed-Resolution Tokens

• URL: http://arxiv.org/abs/2607.26829v1
• Date: Wed, 29 Jul 2026 12:21:40 GMT
• Title: BATS: Resource-Efficient Volumetric Segmentation with Boundary-Aware Mixed-Resolution Tokens
• Abstract summary: We present BATS, a 3D medical image segmentation architecture that concentrates fine-resolution processing near predicted class boundaries.<n>A dense boundary predictor identifies where additional resolution is needed, while a fine-first context constructs an input-dependent mixed-resolution hierarchy.<n>Inference is up to 30% faster on KiTS and LiTS, which retain fewer tokens, but slower on the more token-dense BraTS.
• Score: 14.29921504001782
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Many high-performing volumetric segmentation models maintain dense multi-scale feature maps, leading to high activation memory and inference cost. We present BATS (Boundary-Aware Token Selection), a 3D medical image segmentation architecture that concentrates fine-resolution processing near predicted class boundaries. A dense boundary predictor identifies where additional resolution is needed, while a fine-first context cascade constructs an input-dependent mixed-resolution hierarchy. Homogeneous regions are represented coarsely, with finer tokens retained around boundaries, thin structures, and small targets. The sparse hierarchy is refined and rasterised into a dense segmentation. BATS predicts boundary relevance independently at every resolution level, preventing an erroneous coarse-scale decision from suppressing fine-scale evidence. Parent cluster attention further injects hierarchical ancestor tokens into local attention neighbourhoods, providing cross-scale context without dense multi-scale feature maps or cross-scale neighbour search. We evaluate BATS on five public CT and MRI datasets using the standardised nnU-Net Revisited protocol. BATS achieves the highest LiTS Dice among the compared methods and averages within 0.37 Dice points of the strongest dense baseline, MedNeXt-L, across the five datasets. Relative to MedNeXt-L, it reduces peak allocated GPU memory by more than 53% on KiTS, LiTS, and BraTS. Inference is up to 30% faster on KiTS and LiTS, which retain fewer tokens, but slower on the more token-dense BraTS. Mixed-resolution processing therefore provides consistent memory savings, while runtime and accuracy gains depend on dataset boundary density.

Related papers

• Sparse Attention for Dense Open-Vocabulary Prediction in CLIP [54.45759517856271]
  We study an inference-time substitution of the row-wise softmax in the final visual self-attention layers.<n>Because entmax applies a data-dependent threshold that maps low scores exactly to zero, it acts as an implicit denoiser.<n>We evaluate on open-vocabulary tasks-dense semantic segmentation (Pascal VOC, Pascal Context, ADE20K) and fine-grained retrieval (FG-OVD)
  arXiv  Detail & Related papers  (2026-07-08T08:30:48Z)
• SEMIR: Semantic Minor-Induced Representation Learning on Graphs for Visual Segmentation [1.5222045235700188]
  We introduce SEMIR (Semantic Minor-Induced Representation Learning), a representation framework that decouples inference from the native grid.<n> SEMIR transforms the underlying grid graph into a compact, boundary-aligned graph minor through parameterized edge contraction, node deletion, and edge deletion.<n>We benchmark SEMIR on three tumor segmentation datasets -- BraTS 2021, KiTS23, and LiTS -- where targets exhibit high structural variability and distributional uncertainty.
  arXiv  Detail & Related papers  (2026-05-12T16:52:02Z)
• Binary-Gaussian: Compact and Progressive Representation for 3D Gaussian Segmentation [83.90109373769614]
  3D Gaussian Splatting (3D-GS) has emerged as an efficient 3D representation and a promising foundation for semantic tasks like segmentation.<n>We propose a coarse-to-fine binary encoding scheme for per-Gaussian category representation, which compresses each feature into a single integer via the binary-to-decimal mapping.<n>We further design a progressive training strategy that decomposes panoptic segmentation into a series of independent sub-tasks, reducing inter-class conflicts and thereby enhancing fine-grained segmentation capability.
  arXiv  Detail & Related papers  (2025-11-30T15:51:30Z)
• COB-GS: Clear Object Boundaries in 3DGS Segmentation Based on Boundary-Adaptive Gaussian Splitting [67.03992455145325]
  3D segmentation based on 3D Gaussian Splatting (3DGS) struggles with accurately delineating object boundaries.<n>We introduce Clear Object Boundaries for 3DGS (COB-GS), which aims to improve segmentation accuracy.<n>For semantic guidance, we introduce a boundary-adaptive Gaussian splitting technique.<n>For the visual optimization, we rectify the degraded texture of the 3DGS scene.
  arXiv  Detail & Related papers  (2025-03-25T08:31:43Z)
• Improved distinct bone segmentation in upper-body CT through multi-resolution networks [0.39583175274885335]
  In distinct bone segmentation from upper body CTs a large field of view and a computationally taxing 3D architecture are required. This leads to low-resolution results lacking detail or localisation errors due to missing spatial context. We propose end-to-end trainable segmentation networks that combine several 3D U-Nets working at different resolutions.
  arXiv  Detail & Related papers  (2023-01-31T14:46:16Z)
• Push-the-Boundary: Boundary-aware Feature Propagation for Semantic Segmentation of 3D Point Clouds [0.5249805590164901]
  We propose a boundary-aware feature propagation mechanism to improve semantic segmentation near object boundaries. With one shared encoder, our network outputs (i) boundary localization, (ii) prediction of directions pointing to the object's interior, and (iii) semantic segmentation, in three parallel streams. Our proposed approach yields consistent improvements by reducing boundary errors.
  arXiv  Detail & Related papers  (2022-12-23T15:42:01Z)
• HDNet: A Hierarchically Decoupled Network for Crowd Counting [11.530565995318696]
  We propose a Hierarchically Decoupled Network (HDNet) to solve the above two problems within a unified framework. HDNet achieves state-of-the-art performance on several popular counting benchmarks.
  arXiv  Detail & Related papers  (2022-12-12T06:01:26Z)
• Redesigning Multi-Scale Neural Network for Crowd Counting [68.674652984003]
  We introduce a hierarchical mixture of density experts, which hierarchically merges multi-scale density maps for crowd counting. Within the hierarchical structure, an expert competition and collaboration scheme is presented to encourage contributions from all scales. Experiments show that our method achieves the state-of-the-art performance on five public datasets.
  arXiv  Detail & Related papers  (2022-08-04T21:49:29Z)
• S3Net: 3D LiDAR Sparse Semantic Segmentation Network [1.330528227599978]
  S3Net is a novel convolutional neural network for LiDAR point cloud semantic segmentation. It adopts an encoder-decoder backbone that consists of Sparse Intra-channel Attention Module (SIntraAM) and Sparse Inter-channel Attention Module (SInterAM)
  arXiv  Detail & Related papers  (2021-03-15T22:15:24Z)
• Regularized Densely-connected Pyramid Network for Salient Instance Segmentation [73.17802158095813]
  We propose a new pipeline for end-to-end salient instance segmentation (SIS) To better use the rich feature hierarchies in deep networks, we propose the regularized dense connections. A novel multi-level RoIAlign based decoder is introduced to adaptively aggregate multi-level features for better mask predictions.
  arXiv  Detail & Related papers  (2020-08-28T00:13:30Z)
• Saliency Enhancement using Gradient Domain Edges Merging [65.90255950853674]
  We develop a method to merge the edges with the saliency maps to improve the performance of the saliency. This leads to our proposed saliency enhancement using edges (SEE) with an average improvement of at least 3.4 times higher on the DUT-OMRON dataset. The SEE algorithm is split into 2 parts, SEE-Pre for preprocessing and SEE-Post pour postprocessing.
  arXiv  Detail & Related papers  (2020-02-11T14:04:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.