Diffusion Model-Based Data Augmentation for Enhanced Neuron Segmentation

Hua Han; Jinyue Guo; Liuyun Jiang; Ruining Zhou; Yanchao Zhang; Yizhuo Lu

REVIEW 3 major objections 6 minor 23 references

Synthetic EM images cut neuron-segmentation errors by roughly 30 percent.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 08:45 UTC pith:C7QYCEKH

load-bearing objection A useful diffusion-augmentation recipe for EM neuron segmentation, with a genuine low-annotation win—but the fairness of the headline comparison depends on a data-budget ambiguity the paper never resolves. the 3 major comments →

arxiv 2601.15779 v1 pith:C7QYCEKH submitted 2026-01-22 cs.CV

Diffusion Model-Based Data Augmentation for Enhanced Neuron Segmentation

Liuyun Jiang , Yanchao Zhang , Jinyue Guo , Yizhuo Lu , Ruining Zhou , Hua Han This is my paper

classification cs.CV

keywords electron microscopyneuron segmentationdiffusion modelsdata augmentationconditional generationconnectomicsimage synthesislow-annotation learning

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that a conditional diffusion model can generate realistic 3D electron-microscopy images from neuron labels, and that using these synthetic image-label pairs as training data substantially improves neuron segmentation when manual annotations are scarce. On the AC3/AC4 datasets, augmenting a 4% labeled subset with generated data improves the ARAND error metric by 32.1% and 30.7% under two post-processing pipelines. The key is that the generator is conditioned on multi-scale masks and resolution priors, and that masks are remodeled to be biologically plausible before synthesis. If correct, this offers a path to reduce the heavy annotation burden in connectomics reconstruction.

Core claim

The paper's central claim is that diffusion-based augmentation can turn a small set of manually annotated EM images into a much larger, structurally diverse training set, and that this yields large gains in neuron segmentation accuracy. Specifically, when only 4% of the training volume is labeled, adding generated image-label pairs improves the ARAND metric by 32.1% with Waterz post-processing and 30.7% with Multicut post-processing, relative to training on the real labels alone. The improvement persists across three different segmentation architectures, suggesting the augmentation is not tied to one model family.

What carries the argument

The framework has two parts. First, a resolution-aware conditional diffusion model that synthesizes EM image volumes from 3D masks. It uses multi-scale conditioning (MSC) to inject mask information at every U-Net resolution and a Resolution-Prior Guided Global Modeling (RGM) module that uses Mamba, a state-space model, to capture long-range voxel dependencies while incorporating anisotropic voxel resolutions. Second, a biology-guided mask remodeling module that elastically deforms neuronal membranes and places mitochondria from a signature library into candidate neurons, producing masks that are more realistic than the originals. These remodeled masks are then used as conditions to generate

Load-bearing premise

The 4% labeled subset of the AC4 volume is representative of the whole volume and the reported single-run metrics are stable enough that the 32.1% and 30.7% ARAND improvements are reliable.

What would settle it

If the 4% subset is replaced by a random 4% draw from the full AC4 volume and the reported ARAND gains shrink below statistical significance, or if re-running the same experiment with different random seeds yields a spread larger than the claimed improvements, the central claim would be weakened. A concrete test: run the pipeline on five different 4% subsamples and ten seeds, and report the distribution of ARAND changes.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

If the method holds, connectomics projects could rely on 4% manual labeling plus synthetic data, drastically cutting annotation cost.
The improvement transfers across segmentation backbones (CNN, Transformer, Mamba), so the augmentation is broadly applicable.
The generated data also improves full-annotation training, hinting that synthetic diversity helps even when labels are abundant.
Conditioning on mitochondrial membranes suggests that modeling subcellular structures is important for generating realistic neuron images.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The paper's evaluation uses a single 4% subset and reports no variance across seeds; a more rigorous test would replicate the split and report error bars, and the true gain might be smaller or larger.
The method could be extended to other volumetric segmentation tasks with scarce annotations, such as organelle or vascular structure segmentation, by adapting the mask-remodeling priors.
The 1:1 ratio of generated to real data is a design choice; the optimal ratio may depend on the amount of real data, and a systematic study could push gains further.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

A useful diffusion-augmentation recipe for EM neuron segmentation, with a genuine low-annotation win—but the fairness of the headline comparison depends on a data-budget ambiguity the paper never resolves.

read the letter

You should know two things about this paper before reading it. First, it delivers what it promises: a conditional diffusion pipeline (resolution-aware 3D DDPM with Mamba + multi-scale conditioning) plus a biology-guided mask remodeling step that produces synthetic EM image–mask pairs, and the augmentation consistently improves segmentation across two post-processors and three backbones. That is real, reproducible work, and the code is public. Second, the headline 32.1%/30.7% ARAND improvements at 4% annotation carry a caveat that the paper never clears up, and it is not a minor footnote.

The new stuff: the RPGE encoding that bakes in anisotropic EM resolution, the selective elastic deformation of membranes, and the mitochondrial-placement heuristic. These are domain-specific choices that show up in the ablations as actual contributors. The 3D-FID improvement over Pix2Pix and Med-DDPM is modest but consistent, and the segmentation gains are consistent enough across three architectures that I don't doubt the method helps. The paper is also honest in its structure: ablations for each module, two post-processing pipelines, and a clear statement that all segmentation models were trained from scratch.

The soft spot is in Section 3.3, 'Generated image quality.' That subsection says the generative model was trained on 80 EM images from AC4. No separate protocol is given for the generator used in the augmentation experiments. If that same 80-image-trained generator produced the 1:1 augmentation set for the 4% annotation condition, then the comparison is not apples-to-apples: the baseline segmentation model saw 4 labeled pairs while the augmented model, via the generator, effectively consumed 80. The headline improvement would then be '80 labels beat 4 labels,' which is not what the abstract claims. If instead the generator was retrained on 4 or 20 pairs, the paper must say so—and a diffusion model trained on 4 samples producing useful samples would itself be surprising and worth reporting. This ambiguity is load-bearing, not a nitpick.

Minor complaints: single-run metrics with no standard deviations, no description of how the 4%/20% subsets were sampled, and only two generative baselines for 'state of the art.' These are real but secondary. The core idea and the empirical trend survive the ambiguity; they just cannot be trusted at face value until the data budget is clarified.

Who is this for? People working on EM connectomics or 3D medical image synthesis who want a concrete augmentation recipe. The paper deserves a serious referee, not a desk reject, but the referee should demand the generative data budget, multi-seed variance, and subset sampling details. If the authors clarify the 80-vs-4 point and it turns out the generator was trained only on the same low-annotation budget, this becomes a solid acceptance. As written, it is a conditional.

Referee Report

3 major / 6 minor

Summary. The paper proposes NeuroDiff, a conditional diffusion model for EM neuron segmentation data augmentation. It combines a multi-scale conditioning module (MSC), a Mamba-based global modeling block with resolution-prior encoding (RGM), and a biology-guided mask remodeling module. The authors train a segmentation model on a low-annotation subset of AC4 plus generated image-label pairs, and evaluate on AC3. They report consistent VI/ARAND improvements over unaugmented training, with headline 32.1%/30.7% relative ARAND reductions at 4% annotation, and they study two post-processors and three segmentation backbones.

Significance. The underlying question—whether synthetic image-label pairs can substitute for a large fraction of manual annotations in EM connectomics—is timely and practically important. The architecture choices are well motivated: resolution priors for anisotropic EM volumes, Mamba for global 3D context, and explicit modeling of mitochondria and membrane morphology. The experimental design covers multiple backbones and post-processors, and the ablations in Tables 4 and 5 support the contributions of the MSC, RGM, and mask-remodeling modules. If the gains are reproducible and the labeling budget used to train the generator is clarified, this would be a useful contribution to low-annotation neuron segmentation. However, a critical ambiguity remains about how many labeled pairs the generator consumed, and all quantitative results are single-run point estimates, so the headline numbers are not yet fully interpretable.

major comments (3)

[§3.3 (Generated image quality) and Table 1] The only statement about the generator's training set is: 'We trained the generative model on 80 EM images of size 512×512 from the AC4 dataset and generated samples for the remaining 20 slices.' This appears under 'Generated image quality,' but no separate protocol is stated for the generator used in the Table 1 segmentation-augmentation experiments. If the same generator (trained on 80 AC4 images) is applied under the 4% or 20% annotation condition, then the segmentation model augmented with 'Ours' effectively has access to 80 labeled pairs through the generator, while the 'Real images' baseline sees only 4% or 20% of the labeled data. The 32.1%/30.7% ARAND improvements would then conflate diffusion augmentation with a much larger labeling budget. Conversely, if the generator is retrained on the same 4%/20% subset as the segmentation model, that must be stated; training a diffusion mod
[§3.1 (Datasets and Metrics) and Tables 1,3–5] All quantitative results are single-run point estimates. No standard deviations, confidence intervals, multiple random seeds, or significance tests are reported. Since the headline claims are relative ARAND reductions (e.g., 0.209→0.142 in Table 1), the reader cannot assess whether the gains exceed run-to-run noise, especially under 4% annotation where training is highly stochastic. Additionally, the selection procedure for the 4%/20% labeled subsets is not described: is it random sampling of slices or patches, stratified by neuron density, or a fixed split? Without this, the results are hard to reproduce and the 'low-annotation' conditions are not precisely specified. Please repeat the key experiments with at least 3 random subset draws/seeds and report mean±std, and state the subset sampling protocol.
[§3.2 (Implementation Details) and Table 1] The phrase '4% of the available labeled data' is ambiguous. AC4 has 100 slices, but training is performed on 8×512×512 patches. It is unclear whether 4% means 4 slices, 4% of all possible training patches, or something else. The number of labeled image–mask pairs used in each condition is never given, making the 1:1 augmentation ratio uninterpretable. Please report the actual counts of real and synthetic pairs per condition, so that the reader can determine the effective labeling budget of each method.

minor comments (6)

[Table 3 caption] The text says these results are obtained with a 20% annotation ratio, but the table caption does not state this. Please include the annotation ratio in the caption.
[Table 4] In the ablation rows with a single checkmark, it is not clear which module (MSC or RGM) is being ablated. Label the rows explicitly, e.g., 'MSC only' and 'RGM only'.
[Eq. (3)] Specify whether the coordinates (z_i, y_i, x_i) are normalized or absolute, and how the MLP in RPGE handles volumes of different sizes. The definitions of r_z and r_xy in physical units would also be clearer written as (r_z, r_xy, r_xy).
[Figure 4] Please add scale bars to the generated-image visualizations; without them it is difficult to judge the fidelity and resolution of the synthesized EM images.
[Table 1] In the 100% row, '1.1160.111' should be '1.116 0.111' (missing space).
[General] The paper motivates the approach by the limitations of geometric/photometric augmentation, but no conventional augmentation baseline (e.g., rotation/flip/elastic) is included in Table 1. Adding such a baseline would directly support the claim that diffusion-based augmentation adds value beyond standard augmentations.

Circularity Check

0 steps flagged

No circularity: segmentation gains are measured on held-out AC3 with standard metrics and a from-scratch segmentation model.

full rationale

The central claim—that diffusion-generated image-label pairs improve AC3/AC4 neuron segmentation under low-annotation regimes—is evaluated on a held-out AC3 test set (the first 100 slices) with standard VI and ARAND metrics, and the segmentation model is trained from scratch on mixtures of real and synthetic data. The synthetic data do not define or optimize the evaluation metric, and the reported improvements are arithmetic reductions from Table 1. No fitted parameter is renamed as a prediction, and no load-bearing self-citation chain is present. The auxiliary 3D-FID metric uses a segmentation network pretrained on AC3/AC4 as a feature extractor; this is a domain-fitted feature space, but it does not enter the segmentation benchmark and is not the paper's central claim. The main unresolved ambiguity—whether the generator used for Table 1 was trained on 80 AC4 images while the segmentation model saw only 4% labels—is a data-comparability and possible label-access question, not circularity by construction. The derivation chain is self-contained with respect to the claimed prediction.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 0 invented entities

The central claim is empirical and rests on standard ML training assumptions. The main hand-chosen parameters are mask-remodeling thresholds and the augmentation ratio; the axioms are biological/domain priors about mitochondria, membrane deformation, and distribution shift. No invented entities are introduced.

free parameters (5)

top-10% neuron volume threshold = 10%
Section 2.2: only the top 10% of neurons by volume are candidates for mitochondrial placement; chosen by hand, no sensitivity analysis.
mitochondria axis-length ratio
Section 2.2: mitochondria are selected based on 'a predefined axis-length ratio relative to the corresponding neurons'; exact ratio is not stated.
augmentation ratio (generated:original) = 1:1
Section 3.3: 'the ratio of augmented to original data was set to 1:1' — hand-chosen, affects training mix.
diffusion training iterations = 10,000
Section 3.2: trained for 10,000 iterations with lr 1e-5, batch size 1; standard hyperparameter but affects generation quality.
segmentation training iterations = 200,000
Section 3.2: Superhuman trained for 200,000 iterations, lr 1e-4, batch size 2; standard but affects final performance.

axioms (5)

domain assumption Mitochondrial membranes affect the identification of neuronal boundaries.
Motivates conditioning on both neuronal boundaries and mitochondrial membranes (Sections 2.1 and 2.2). If false, the conditioning signal is less useful.
domain assumption Elastically deforming only the membrane mask preserves global structural integrity while producing realistic boundary variations.
Assumed in Section 2.2; no quantitative validation of biological plausibility beyond downstream segmentation results.
domain assumption Synthetic images generated from remodeled masks are close enough to the real AC4 distribution to improve training of a segmentation model evaluated on AC3.
Central to the augmentation strategy; tested indirectly through segmentation metrics but not through distribution analysis per annotation ratio.
domain assumption The first 100 slices of AC3 are a representative test set for evaluating neuron segmentation.
Section 3.1: test set is a fixed 100-slice subvolume; no evidence that this region is representative of the full AC3 volume.
standard math Standard DDPM formulation with learned Gaussian transitions is a valid generative model for 3D EM volumes.
Equations (1)-(2) rely on the standard DDPM framework from Ho et al. [7], accepted background.

pith-pipeline@v1.3.0-alltime-deepseek · 6474 in / 10639 out tokens · 151054 ms · 2026-08-03T08:45:35.892222+00:00 · methodology

0 comments

read the original abstract

Neuron segmentation in electron microscopy (EM) aims to reconstruct the complete neuronal connectome; however, current deep learning-based methods are limited by their reliance on large-scale training data and extensive, time-consuming manual annotations. Traditional methods augment the training set through geometric and photometric transformations; however, the generated samples remain highly correlated with the original images and lack structural diversity. To address this limitation, we propose a diffusion-based data augmentation framework capable of generating diverse and structurally plausible image-label pairs for neuron segmentation. Specifically, the framework employs a resolution-aware conditional diffusion model with multi-scale conditioning and EM resolution priors to enable voxel-level image synthesis from 3D masks. It further incorporates a biology-guided mask remodeling module that produces augmented masks with enhanced structural realism. Together, these components effectively enrich the training set and improve segmentation performance. On the AC3 and AC4 datasets under low-annotation regimes, our method improves the ARAND metric by 32.1% and 30.7%, respectively, when combined with two different post-processing methods. Our code is available at https://github.com/HeadLiuYun/NeuroDiff.

discussion (0)

Reference graph

Works this paper leans on

23 extracted references · 5 linked inside Pith

[1]

INTRODUCTION Neuron segmentation aims to elucidate brain function by mapping neural connectivity and analyzing inter-neuronal signaling pathways. Advances in volume EM have enabled nanoscale reconstruction of three-dimensional neuronal struc- tures [1, 2]; however, the increasing data scale and resolution pose significant challenges for automated processi...

Pith/arXiv arXiv 2026
[2]

Resolution-Aware Conditional Diffusion In this study, we employ a Denoising Diffusion Probabilistic Model (DDPM) [7] for 3D EM image synthesis

METHOD 2.1. Resolution-Aware Conditional Diffusion In this study, we employ a Denoising Diffusion Probabilistic Model (DDPM) [7] for 3D EM image synthesis. Following previous 3D diffusion studies [8, 5], we extend the U-Net to a 3D architecture for volumetric generation. In the forward process, Gaussian noiseϵ∼ N(0, I)is gradually added to the datax 0 ove...
[3]

Real im- ages

EXPERIMENTS 3.1. Datasets and Metrics We used the AC3 and AC4 datasets from the mouse so- matosensory cortex [12], imaged using scanning electron microscopy (SEM) at a resolution of6×6×29 nm 3, with Table 1. Comparison of segmentation performance using dif- ferent synthesis methods for data augmentation across vary- ing annotation ratios on the AC3/AC4 da...
[4]

Experiments demonstrate that our method effectively improves neuron segmentation performance

CONCLUSION This paper proposes a diffusion-based data augmentation framework for neuron segmentation, which synthesizes new image–label pairs to enhance model training. Experiments demonstrate that our method effectively improves neuron segmentation performance
[5]

Eth- ical approval was not required as confirmed by the license attached with the open access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using mouse subject data made available in open access [12]. Eth- ical approval was not required as confirmed by the license attached with the open access data
[6]

ACKNOWLEDGEMENTS This work was supported by the Beijing Natural Science Foundation (No. 5254042), the Brain Science and Brain-like Intelligence Technology – National Science and Technology Major Project (2022ZD0211900, 2022ZD0211902), and the National Natural Science Foundation of China (No. 32171461)
[7]

Exploring the third dimension: volume electron microscopy comes of age,

Christopher J Peddie and Lucy M Collinson, “Exploring the third dimension: volume electron microscopy comes of age,”Micron, vol. 61, pp. 9–19, 2014

2014
[8]

V olume electron mi- croscopy,

Christopher J Peddie, Christel Genoud, Anna Kreshuk, Kimberly Meechan, Kristina D Micheva, Kedar Narayan, Constantin Pape, Robert G Parton, Nicole L Schieber, Yannick Schwab, et al., “V olume electron mi- croscopy,”Nature Reviews Methods Primers, vol. 2, no. 1, pp. 51, 2022

2022
[9]

Superhuman accuracy on the snemi3d connectomics challenge,

Kisuk Lee, Jonathan Zung, Peter Li, Viren Jain, and H Sebastian Seung, “Superhuman accuracy on the snemi3d connectomics challenge,”arXiv preprint arXiv:1706.00120, 2017

Pith/arXiv arXiv 2017
[10]

Mask rearranging data augmentation for 3d mi- tochondria segmentation,

Qi Chen, Mingxing Li, Jiacheng Li, Bo Hu, and Zhiwei Xiong, “Mask rearranging data augmentation for 3d mi- tochondria segmentation,” inInternational Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2022, pp. 36–46

2022
[11]

Conditional diffusion models for semantic 3d medical image synthesis,

Zolnamar Dorjsembe, Hsing-Kuo Pao, Sodtavilan Odonchimed, and Furen Xiao, “Conditional diffusion models for semantic 3d medical image synthesis,”Au- thorea Preprints, 2023

2023
[12]

Red blood cell image generation for data augmentation using conditional generative adversarial networks,

Oleksandr Bailo, DongShik Ham, and Young Min Shin, “Red blood cell image generation for data augmentation using conditional generative adversarial networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, 2019, pp. 0– 0

2019
[13]

Denoising diffusion probabilistic models,

Jonathan Ho, Ajay Jain, and Pieter Abbeel, “Denoising diffusion probabilistic models,”Advances in neural in- formation processing systems, vol. 33, pp. 6840–6851, 2020

2020
[14]

Three-dimensional medical image syn- thesis with denoising diffusion probabilistic models,

Zolnamar Dorjsembe, Sodtavilan Odonchimed, and Furen Xiao, “Three-dimensional medical image syn- thesis with denoising diffusion probabilistic models,” in Medical imaging with deep learning, 2022

2022
[15]

Mamba: Linear-time sequence modeling with selective state spaces,

Albert Gu and Tri Dao, “Mamba: Linear-time sequence modeling with selective state spaces,”arXiv preprint arXiv:2312.00752, 2023

Pith/arXiv arXiv 2023
[16]

A new approach to linear filter- ing and prediction problems,

Rudolph Emil Kalman, “A new approach to linear filter- ing and prediction problems,” 1960

1960
[17]

An image is worth 16x16 words: Transformers for image recognition at scale,

Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al., “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

Pith/arXiv arXiv 2010
[18]

Saturated reconstruction of a volume of neocortex,

Narayanan Kasthuri, Kenneth Jeffrey Hayworth, Daniel Raimund Berger, Richard Lee Schalek, Jos´e An- gel Conchello, Seymour Knowles-Barley, Dongil Lee, Amelio V ´azquez-Reina, Verena Kaynig, Thouis Ray- mond Jones, et al., “Saturated reconstruction of a volume of neocortex,”Cell, vol. 162, no. 3, pp. 648–661, 2015

2015
[19]

Image-to-image translation with conditional ad- versarial networks,

Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros, “Image-to-image translation with conditional ad- versarial networks,” inProceedings of the IEEE confer- ence on computer vision and pattern recognition, 2017, pp. 1125–1134

2017
[20]

Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,

Ali Hatamizadeh, Vishwesh Nath, Yucheng Tang, Dong Yang, Holger R Roth, and Daguang Xu, “Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,” inInternational MICCAI brain- lesion workshop. Springer, 2021, pp. 272–284

2021
[21]

Segmamba: Long-range sequential model- ing mamba for 3d medical image segmentation,

Zhaohu Xing, Tian Ye, Yijun Yang, Guang Liu, and Lei Zhu, “Segmamba: Long-range sequential model- ing mamba for 3d medical image segmentation,”arXiv preprint arXiv:2401.13560, 2024

Pith/arXiv arXiv 2024
[22]

Large scale image segmentation with structured loss based deep learning for connectome re- construction,

Jan Funke, Fabian Tschopp, William Grisaitis, Arlo Sheridan, Chandan Singh, Stephan Saalfeld, and Srini- vas C Turaga, “Large scale image segmentation with structured loss based deep learning for connectome re- construction,”IEEE transactions on pattern analysis and machine intelligence, vol. 41, no. 7, pp. 1669–1680, 2018

2018
[23]

Multicut brings automated neurite seg- mentation closer to human performance,

Thorsten Beier, Constantin Pape, Nasim Rahaman, Timo Prange, Stuart Berg, Davi D Bock, Albert Car- dona, Graham W Knott, Stephen M Plaza, Louis K Scheffer, et al., “Multicut brings automated neurite seg- mentation closer to human performance,”Nature meth- ods, vol. 14, no. 2, pp. 101–102, 2017

2017

[1] [1]

INTRODUCTION Neuron segmentation aims to elucidate brain function by mapping neural connectivity and analyzing inter-neuronal signaling pathways. Advances in volume EM have enabled nanoscale reconstruction of three-dimensional neuronal struc- tures [1, 2]; however, the increasing data scale and resolution pose significant challenges for automated processi...

Pith/arXiv arXiv 2026

[2] [2]

Resolution-Aware Conditional Diffusion In this study, we employ a Denoising Diffusion Probabilistic Model (DDPM) [7] for 3D EM image synthesis

METHOD 2.1. Resolution-Aware Conditional Diffusion In this study, we employ a Denoising Diffusion Probabilistic Model (DDPM) [7] for 3D EM image synthesis. Following previous 3D diffusion studies [8, 5], we extend the U-Net to a 3D architecture for volumetric generation. In the forward process, Gaussian noiseϵ∼ N(0, I)is gradually added to the datax 0 ove...

[3] [3]

Real im- ages

EXPERIMENTS 3.1. Datasets and Metrics We used the AC3 and AC4 datasets from the mouse so- matosensory cortex [12], imaged using scanning electron microscopy (SEM) at a resolution of6×6×29 nm 3, with Table 1. Comparison of segmentation performance using dif- ferent synthesis methods for data augmentation across vary- ing annotation ratios on the AC3/AC4 da...

[4] [4]

Experiments demonstrate that our method effectively improves neuron segmentation performance

CONCLUSION This paper proposes a diffusion-based data augmentation framework for neuron segmentation, which synthesizes new image–label pairs to enhance model training. Experiments demonstrate that our method effectively improves neuron segmentation performance

[5] [5]

Eth- ical approval was not required as confirmed by the license attached with the open access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using mouse subject data made available in open access [12]. Eth- ical approval was not required as confirmed by the license attached with the open access data

[6] [6]

ACKNOWLEDGEMENTS This work was supported by the Beijing Natural Science Foundation (No. 5254042), the Brain Science and Brain-like Intelligence Technology – National Science and Technology Major Project (2022ZD0211900, 2022ZD0211902), and the National Natural Science Foundation of China (No. 32171461)

[7] [7]

Exploring the third dimension: volume electron microscopy comes of age,

Christopher J Peddie and Lucy M Collinson, “Exploring the third dimension: volume electron microscopy comes of age,”Micron, vol. 61, pp. 9–19, 2014

2014

[8] [8]

V olume electron mi- croscopy,

Christopher J Peddie, Christel Genoud, Anna Kreshuk, Kimberly Meechan, Kristina D Micheva, Kedar Narayan, Constantin Pape, Robert G Parton, Nicole L Schieber, Yannick Schwab, et al., “V olume electron mi- croscopy,”Nature Reviews Methods Primers, vol. 2, no. 1, pp. 51, 2022

2022

[9] [9]

Superhuman accuracy on the snemi3d connectomics challenge,

Kisuk Lee, Jonathan Zung, Peter Li, Viren Jain, and H Sebastian Seung, “Superhuman accuracy on the snemi3d connectomics challenge,”arXiv preprint arXiv:1706.00120, 2017

Pith/arXiv arXiv 2017

[10] [10]

Mask rearranging data augmentation for 3d mi- tochondria segmentation,

Qi Chen, Mingxing Li, Jiacheng Li, Bo Hu, and Zhiwei Xiong, “Mask rearranging data augmentation for 3d mi- tochondria segmentation,” inInternational Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2022, pp. 36–46

2022

[11] [11]

Conditional diffusion models for semantic 3d medical image synthesis,

Zolnamar Dorjsembe, Hsing-Kuo Pao, Sodtavilan Odonchimed, and Furen Xiao, “Conditional diffusion models for semantic 3d medical image synthesis,”Au- thorea Preprints, 2023

2023

[12] [12]

Red blood cell image generation for data augmentation using conditional generative adversarial networks,

Oleksandr Bailo, DongShik Ham, and Young Min Shin, “Red blood cell image generation for data augmentation using conditional generative adversarial networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, 2019, pp. 0– 0

2019

[13] [13]

Denoising diffusion probabilistic models,

Jonathan Ho, Ajay Jain, and Pieter Abbeel, “Denoising diffusion probabilistic models,”Advances in neural in- formation processing systems, vol. 33, pp. 6840–6851, 2020

2020

[14] [14]

Three-dimensional medical image syn- thesis with denoising diffusion probabilistic models,

Zolnamar Dorjsembe, Sodtavilan Odonchimed, and Furen Xiao, “Three-dimensional medical image syn- thesis with denoising diffusion probabilistic models,” in Medical imaging with deep learning, 2022

2022

[15] [15]

Mamba: Linear-time sequence modeling with selective state spaces,

Albert Gu and Tri Dao, “Mamba: Linear-time sequence modeling with selective state spaces,”arXiv preprint arXiv:2312.00752, 2023

Pith/arXiv arXiv 2023

[16] [16]

A new approach to linear filter- ing and prediction problems,

Rudolph Emil Kalman, “A new approach to linear filter- ing and prediction problems,” 1960

1960

[17] [17]

An image is worth 16x16 words: Transformers for image recognition at scale,

Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al., “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

Pith/arXiv arXiv 2010

[18] [18]

Saturated reconstruction of a volume of neocortex,

Narayanan Kasthuri, Kenneth Jeffrey Hayworth, Daniel Raimund Berger, Richard Lee Schalek, Jos´e An- gel Conchello, Seymour Knowles-Barley, Dongil Lee, Amelio V ´azquez-Reina, Verena Kaynig, Thouis Ray- mond Jones, et al., “Saturated reconstruction of a volume of neocortex,”Cell, vol. 162, no. 3, pp. 648–661, 2015

2015

[19] [19]

Image-to-image translation with conditional ad- versarial networks,

Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros, “Image-to-image translation with conditional ad- versarial networks,” inProceedings of the IEEE confer- ence on computer vision and pattern recognition, 2017, pp. 1125–1134

2017

[20] [20]

Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,

Ali Hatamizadeh, Vishwesh Nath, Yucheng Tang, Dong Yang, Holger R Roth, and Daguang Xu, “Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,” inInternational MICCAI brain- lesion workshop. Springer, 2021, pp. 272–284

2021

[21] [21]

Segmamba: Long-range sequential model- ing mamba for 3d medical image segmentation,

Zhaohu Xing, Tian Ye, Yijun Yang, Guang Liu, and Lei Zhu, “Segmamba: Long-range sequential model- ing mamba for 3d medical image segmentation,”arXiv preprint arXiv:2401.13560, 2024

Pith/arXiv arXiv 2024

[22] [22]

Large scale image segmentation with structured loss based deep learning for connectome re- construction,

Jan Funke, Fabian Tschopp, William Grisaitis, Arlo Sheridan, Chandan Singh, Stephan Saalfeld, and Srini- vas C Turaga, “Large scale image segmentation with structured loss based deep learning for connectome re- construction,”IEEE transactions on pattern analysis and machine intelligence, vol. 41, no. 7, pp. 1669–1680, 2018

2018

[23] [23]

Multicut brings automated neurite seg- mentation closer to human performance,

Thorsten Beier, Constantin Pape, Nasim Rahaman, Timo Prange, Stuart Berg, Davi D Bock, Albert Car- dona, Graham W Knott, Stephen M Plaza, Louis K Scheffer, et al., “Multicut brings automated neurite seg- mentation closer to human performance,”Nature meth- ods, vol. 14, no. 2, pp. 101–102, 2017

2017