---
title: 'HFF-Net: Frequency Fusion for Brain Tumor Segmentation'
url: https://www.emergentmind.com/topics/hff-net
type: topic
---

# HFF-Net: Frequency Fusion for Brain Tumor Segmentation

Searching arXiv for recent papers using the term "HFF-Net" and closely related variants.
Search query: HFF-Net frequency brain tumor segmentation arXiv
HFF-Net most commonly denotes the **Harmonized Frequency Fusion Network**, a frequency-aware 3D brain tumor segmentation architecture that decomposes multi-modal MRI into complementary low-frequency and multi-directional high-frequency representations, processes them in separate branches, and fuses them for subregion delineation [2506.10142]. In later BraTS-PED 2025 work, the same model family is used as the frequency-aware component of an ensemble with nnU-Net and Swin UNETR, where its intended contribution is sharper boundary modeling and improved sensitivity to texture-rich tumor regions in pediatric MRI [2509.19353]. The name is not unique across recent literature, however, and has also been used in unrelated contexts, making terminological disambiguation necessary.

## 1. Terminology and scope

In current arXiv literature, **HFF-Net** primarily refers to a medical image segmentation model built around explicit frequency decomposition of MRI volumes [2506.10142]. That usage is the dominant one in brain tumor segmentation and is the basis of the BraTS-PED 2025 ensemble instantiation [2509.19353].

The term is nevertheless ambiguous. Separate papers use closely related names for different systems and learning paradigms. Only the first two entries below refer to the same segmentation lineage.

| Name in paper | Domain | Source |
|---|---|---|
| HFF-Net | Harmonized Frequency Fusion Network for 3D brain tumor segmentation | [2506.10142] |
| HFF-Net | Frequency-aware branch in a BraTS-PED 2025 ensemble | [2509.19353] |
| HFF-Net / HF-FMs | Hierarchical networked system for federated foundation models over wireless networks | [2509.03695] |
| HFF | Hyperspherical Forward-Forward local learning algorithm | [2605.00082] |
| HFFN | High-Frequency Focused Network for single-image super-resolution | [2303.11701] |

In the segmentation literature, HFF-Net is motivated by the observation that low-frequency MRI content encodes smooth global morphology, whereas high-frequency content captures sharp boundaries, anisotropic textures, and subtle variations that are particularly important for enhancing tumor segmentation [2506.10142]. The BraTS-PED 2025 paper adopts that frequency-aware premise and applies it to pediatric high-grade glioma segmentation, where heterogeneous anatomy, small lesions, and acquisition variability are emphasized as central difficulties [2509.19353].

## 2. Problem setting and clinical target structure

The original HFF-Net paper formulates the method for 3D brain tumor segmentation on multi-modal MRI, with a particular emphasis on contrast-enhancing tumor regions visible in post-contrast T1-weighted MRI [2506.10142]. The datasets listed for evaluation are BraTS2019, BraTS2020, MSD-BTS, and BraTS2023-MEN, using four modalities per case: T1, T1Gd or T1C, T2, and FLAIR or T2-FLAIR, depending on the benchmark.

Its core segmentation targets are the three major BraTS subregions: **ET**, **WT**, and **TC**. In the BraTS-PED 2025 ensemble setting, the label space is expanded to six regions: **ET** (enhancing tumor), **NET** (non-enhancing tumor), **CC** (cystic component), **ED** (peritumoral edema), **TC** (tumor core), and **WT** (whole tumor), with the standard BraTS compositional semantics that $\mathrm{TC} = \mathrm{ET} \cup \mathrm{NET} \cup \mathrm{CC}$ and $\mathrm{WT} = \mathrm{TC} \cup \mathrm{ED}$ [2509.19353].

The frequency-domain rationale is consistent across both papers. Low-frequency components are described as supporting smooth tissue contours and robust global delineation, while high-frequency components emphasize directional edges, fine textures, and boundary detail. In the pediatric setting, this division is explicitly linked to the hypothesis that low-frequency structure helps ED and overall tumor extent, whereas high-frequency structure helps ET and CC characterization [2509.19353]. This suggests that HFF-Net is not merely a preprocessing variant of a conventional encoder-decoder, but a model designed to encode a specific anatomical prior about how tumor evidence is distributed across spectral bands.

## 3. Architecture and frequency-aware representation

HFF-Net is a **dual-branch 3D encoder-decoder** that separately processes low-frequency and high-frequency information and fuses them at deeper stages of the network [2506.10142]. In the full formulation, three named modules define the architecture: **Frequency Domain Decomposition (FDD)**, **Adaptive Laplacian Convolution (ALC)**, and **Frequency Domain Cross-Attention (FDCA)**.

### Frequency Domain Decomposition

FDD decomposes each modality into one low-frequency component and multiple directional high-frequency components. The low-frequency path uses the **Dual-Tree Complex Wavelet Transform (DTCWT)**, while the high-frequency path uses the **Nonsubsampled Contourlet Transform (NSCT)**. In the compact formulation reported for BraTS-PED 2025, for each modality $m$ and slice $s$,
$$
LF_m[s] = \phi_{\mathrm{DTCWT}} * I_m[s], \qquad HF_{m,k}[s] = \psi_{\mathrm{NSCT},k} * I_m[s], \; k \in \{1,2,3,4\}.
$$
Stacking slices reconstructs 3D low-frequency and high-frequency volumes at the original spatial resolution [2509.19353].

In the original HFF-Net paper, the FDD outputs are given as
- $x^L \in \mathbb{R}^{M \times D \times H \times W}$ for low-frequency tensors, and
- $x^H \in \mathbb{R}^{4M \times D \times H \times W}$ for high-frequency tensors,

with four directional high-frequency subbands per modality [2506.10142]. In the BraTS-PED 2025 implementation this becomes **20 input channels per case**: 4 low-frequency channels and 16 high-frequency channels [2509.19353].

### Adaptive Laplacian Convolution

ALC is the mechanism that adapts high-frequency filtering while preserving explicit edge sensitivity. Its kernels are initialized with the discrete 3D Laplacian operator,
$$
(\Delta X)(i,j,k) = X(i+1,j,k)+X(i-1,j,k)+X(i,j+1,k)+X(i,j-1,k)+X(i,j,k+1)+X(i,j,k-1) - 6X(i,j,k),
$$
and are then regularized with an Elastic Weight Consolidation scheme that preserves weights deemed important by Fisher information after a 40-epoch warmup [2506.10142]. Operationally, the paper summarizes the module as
$$
y = \mathrm{Conv}(x; K_{\mathrm{ALC}}), \qquad K_{\mathrm{ALC}} = \mathrm{LaplaceInit} + \Delta K.
$$

This design is intended to preserve high-pass behavior while still adapting to multimodal, multi-directional data. The paper further reports that the discrete Laplacian outperforms Sobel, Scharr, and Kirsch initializations in its operator-sensitivity study [2506.10142].

### Frequency Domain Cross-Attention

FDCA performs cross-frequency fusion in the frequency domain using **semantic**, **positional**, and **slice-specific** attentions. For feature maps $F \in \mathbb{R}^{c \times n \times h \times w}$, FFT is applied to obtain real and imaginary components, and three attention stages are computed in sequence. The module is described as integrating semantic, positional, and slice-specific information to improve anisotropic volumetric fusion and cross-frequency consistency [2506.10142].

The slice-attention mechanism is particularly specific to the 3D setting: it operates along the depth axis $n$, and the paper states that it is intended to simulate inter-slice variability and maintain volumetric coherence. This distinguishes HFF-Net from purely spectral or purely spatial attention schemes.

### Encoder-decoder organization

The original paper describes dedicated LF and HF encoders, deep fusion, and a symmetric decoder with skip connections and side-output supervision [2506.10142]. The BraTS-PED 2025 summary, which relies on the previously introduced HFF-Net, describes the model more generically as a standard 3D encoder-decoder with skip connections in both LF and HF branches and fusion near the bottleneck or decoder, while noting that exact layer counts, kernel sizes, channel widths, and module details are not disclosed there [2509.19353].

## 4. Training objectives, normalization, and optimization

HFF-Net is trained on **3D patches of size $128 \times 128 \times 128$** and applies instance-level or volume-level z-score normalization,
$$
I' = \frac{I - \mu}{\sigma}.
$$
This normalization is explicitly reported in the BraTS-PED 2025 implementation and is also consistent with the preprocessing described in the original paper [2509.19353].

In the full HFF-Net formulation, the optimization target is
$$
\mathcal{L}_{\mathrm{total}} = \mathcal{L}_{\mathrm{sup}} + \lambda_1 \mathcal{L}_{\mathrm{unsup}} + \lambda_2 \mathcal{L}_{\mathrm{ewc}},
$$
where supervised segmentation uses Dice loss, unsupervised LF-HF consistency is enforced through a **3D Dynamic Focal Loss (DFL)** in the frequency domain, and $\mathcal{L}_{\mathrm{ewc}}$ constrains ALC kernels relative to the Laplacian prior [2506.10142]. The supervised Dice formulation is
$$
\mathcal{L}_{\mathrm{Dice}} = 1 - \frac{2\sum_i p_i g_i + \epsilon}{\sum_i p_i + \sum_i g_i + \epsilon}.
$$

The original training protocol uses **SGD**, momentum **0.9**, weight decay **$5 \times 10^{-5}$**, initial learning rate **0.3**, multiplicative decay by **0.53 every 50 epochs**, total **350 epochs**, batch size **1**, and a **40-epoch warmup** for EWC and the ramping of $\lambda_1$ to $\lambda_{\max}=15$ [2506.10142]. Augmentations are random flipping along each axis, rotations within $\pm 10^\circ$, and random crops centered around the brain.

The BraTS-PED 2025 paper reports a closely related but not identical schedule for its HFF-Net component: **450 epochs**, batch size **1**, patch size **$128 \times 128 \times 128$**, **SGD** with momentum **0.9**, weight decay **$5 \times 10^{-5}$**, and initial learning rate **0.3** with progressive decay; mixed precision, HFF-Net-specific augmentations beyond z-score normalization, and the exact loss are not stated there [2509.19353]. That paper also explicitly notes that, unlike Swin UNETR, HFF-Net is trained from scratch in the pediatric setting.

## 5. Empirical results and role within ensembles

On the multi-dataset evaluation reported for the original model, HFF-Net achieves the following representative Dice and HD95 results [2506.10142]:

- **BraTS2023-MEN**: Dice ET **96.16%**, WT **95.56%**, TC **96.34%**; HD95 ET **5.576 mm**, WT **6.272 mm**, TC **6.432 mm**.
- **MSD-BTS**: Dice ET **87.28%**, WT **94.03%**, TC **89.09%**.
- **BraTS2020**: Dice ET **87.36%**, WT **92.37%**, TC **88.22%**.
- **BraTS2019**: Dice ET **85.35%**, WT **91.58%**, TC **87.59%**.

The abstract reports an **average relative improvement of 4.48\%** in mean Dice across ET, WT, and TC, with a range of **2.39\% to 7.72\%**, and an **average relative improvement of 7.33\%** in ET Dice, with a range of **5.96\% to 8.64\%** [2506.10142]. The paper also states that HFF-Net shows significantly higher Dice and lower HD95, with **$p < 0.05$** in multiple comparisons against baselines including UNETR, VT-UNet, PANet, SegMamba, UNETR++, and nnU-Net, especially for ET.

Ablation results support the architectural decomposition. On BraTS23-MEN, **FDD alone** yields ET **92.89**, WT **91.38**, TC **92.48**; **FDD + ALC** improves ET by **+1.96** to **95.89**; **FDD + FDCA** raises TC to **95.56**; and the **full model** reaches ET **96.16**, WT **95.56**, TC **96.34**, with **~37.76M** parameters and **~573.06 GFLOPs** [2506.10142].

In the BraTS-PED 2025 ensemble study, standalone HFF-Net validation performance is reported as Dice **0.683** for CC, **0.934** for ED, **0.703** for ET, **0.900** for NET, **0.928** for TC, and **0.928** for WT, with corresponding **NSD-0.5** values of **0.673**, **0.934**, **0.638**, **0.653**, **0.674**, and **0.675** [2509.19353]. The paper states that HFF-Net’s strongest standalone improvement appears on ET, which it associates with the model’s emphasis on boundaries and directional texture.

The final pediatric ensemble uses equal-weight probability averaging,
$$
S_{\mathrm{ens}} = \frac{1}{3} S_{\mathrm{nnUNet}} + \frac{1}{3} S_{\mathrm{SwinUNETR}} + \frac{1}{3} S_{\mathrm{HFFNet}},
$$
followed by argmax, with no post-processing [2509.19353]. The reported challenge summary scores are **72.3%** for ET, **95.6%** for NET, **68.9%** for CC, **89.5%** for ED, **92.3%** for TC, and **92.3%** for WT. On the validation table, the ensemble reaches ET **0.689**, CC **0.723**, ED **0.956**, NET **0.895**, TC **0.923**, and WT **0.923**.

## 6. Computational profile, reproducibility, and implementation

The original HFF-Net study reports implementation in **PyTorch 2.1.2** with **CUDA 11.8**, training on an **NVIDIA RTX 4090 (24 GB)** [2506.10142]. The reported full-model computational profile includes around **36.01M parameters** and **~541.18 GFLOPs** in the BraTS2020 table, increasing to **~37.76M** parameters and **~573.06 GFLOPs** in ablations. Reported inference performance is **~6.71 images/s** with **~159.3 ms/image** GPU latency.

The BraTS-PED 2025 ensemble implementation reports **NVIDIA GeForce RTX 4080 (16 GB)** hardware and an HFF-Net training time of **≈48 hours** [2509.19353]. For comparison within the same paper, a single-fold nnU-Net requires **≈37 hours**, and Swin UNETR requires **≈4 days**. Parameter counts, memory footprint, and inference-time latency are not reported there.

The reproducibility path is unusually explicit at the preprocessing level. The pediatric paper specifies:
1. skull stripping via a public nnU-Net-based pipeline,
2. per-modality frequency decomposition through **DTCWT** for LF and **NSCT** for four directional HF subbands,
3. reassembly into 3D **20-channel** inputs,
4. z-score normalization of each channel volume,
5. training on **$128^3$** patches,
6. equal-weight probability averaging with nnU-Net and Swin UNETR [2509.19353].

Both the original HFF-Net study and the pediatric ensemble reference the same code repository: **https://github.com/VinyehShaw/HFF** [2506.10142]. The pediatric paper additionally notes that no post-processing such as connected-component filtering, hole filling, or region-size filtering is applied [2509.19353].

## 7. Limitations, open questions, and related misconceptions

A persistent limitation is that **ET remains the most challenging class** even with frequency-aware modeling in the pediatric setting, which the paper attributes to small, irregular, and heterogeneous enhancement patterns [2509.19353]. The original study also notes failure cases in which missing or corrupted FLAIR impairs the separation of ED from NCR/NET, indicating that decomposition alone does not neutralize modality unreliability [2506.10142].

Another limitation is that the frequency transforms used in the BraTS-PED configuration are **fixed** rather than learned. The paper explicitly remarks that robustness to acquisition variability may therefore be limited, because the network cannot adapt cutoffs or orientations to scanner-dependent artifacts [2509.19353]. The full HFF-Net paper likewise acknowledges the computational burden of the model relative to lightweight baselines, despite reporting favorable clinical applicability [2506.10142].

The future directions named across the two papers are technically coherent. They include **adaptive frequency masks**, **learned cutoffs and orientations**, **learnable wavelet- or contourlet-like convolutions**, **multi-scale frequency fusion**, **wavelet packet or hybrid spatial-frequency transformers**, and **lightweight post-processing** for ET and CC suppression of spurious high-frequency false positives [2509.19353]. The original paper also motivates further work on preserving more spatial information beyond the current FDCA and decoder design, while maintaining the clinical advantages of end-to-end inference without post-processing [2506.10142].

A common misconception is to treat HFF-Net as a generic “frequency-domain CNN.” The published description is narrower and more structured than that. It is specifically a **3D dual-branch encoder-decoder** whose spectral inductive bias is implemented through DTCWT-based LF decomposition, NSCT-based directional HF decomposition, Laplacian-initialized adaptive high-pass filtering, and frequency-domain cross-attention [2506.10142]. A second misconception is terminological: not every paper using a similar acronym refers to this segmentation model. The wireless networking literature’s HFF-Net is a hierarchical orchestration fabric for federated foundation models, not a medical segmentation architecture [2509.03695], and HFF is also used for a hyperspherical reformulation of the Forward-Forward learning algorithm [2605.00082].

Source: https://www.emergentmind.com/topics/hff-net