---
格式版本: 2
标题: "Automated 2D and 3D Segmentation of AMD and DME Lesions in OCT"
原文链接: "https://arxiv.org/abs/2608.27095"
发布日期: "2026-08-27"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Thu, 27 Aug 2026 13:13:18 UTC (15,326 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-30T15:19:36+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-30T15:19:02+08:00"
入库时间: "2026-08-30T07:19:36.561Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=AMD&searchtype=all"
匹配关键词:
  []
相关厂家:
  - "AMD"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 23
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线是AMD（年龄相关性黄斑变性）与DME眼科病灶的OCT图像分割研究，并非AMD公司或机架级AI基础设施。当前来源为可追溯的arXiv预印本摘要页，披露了分割流程、Dice及校准指标，但没有超节点架构、互连、供电、液冷、RAS或商业部署信息。固定知识库未显示同项历史成果，但其新增事实仅属于医疗视觉模型评估。命中“应用与模型效率”强否决项，不能进入业务情报库。"
AI质检模型: "gpt-5.6-sol"
AI质检时间: "2026-08-30T15:19:49+08:00"
AI主题相关性: 0
AI来源权威性: 8
AI新颖性: 7
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 8
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.55"
AI评分知识库SHA256: "ca56afec1ff616b1fa4394cd1d08b1e2a6af47eccbe70c802c9f1fea6037e803"
AI评分知识库检索词: "[\"AMD\",\"DME\",\"OCT\",\"PDF\",\"arxiv.org/pdf/2608.27095\",\"HTML\",\"arxiv.org/html/2608.27095v1\",\"OLIVES\",\"AUROC\",\"CST\",\"arxiv.org/abs/2608.27095\",\"arxiv.org/abs/2608.27095v1\"]"
AI评分知识库命中: "[{\"id\":\"july-correct-0081\",\"title\":\"AAI 2026: 6th Gen AMD EPYC Server CPUs Power the Agentic Data Center\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"AMD\",\"PDF\",\"HTML\"],\"rank\":-8.29317668336406},{\"id\":\"runtime-569639751a0dbe7ef3ffccc5\",\"title\":\"[2608.17503] Predict Before Replay: Joint FEC and Flight Control for Reliable Scale-Up Links\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-18\",\"matchedTerms\":[\"PDF\",\"HTML\"],\"rank\":-8.202857281562261},{\"id\":\"july-correct-0010\",\"title\":\"Schneider Electric and AMD release first Helios platform reference design to accelerate AI Factory deployment\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"AMD\",\"OCT\",\"PDF\"],\"rank\":-7.446429491258733},{\"id\":\"july-correct-0111\",\"title\":\"StrataCL: Fabric-Native Communication Library for Production Supernodes\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"PDF\",\"HTML\"],\"rank\":-6.753387360805314},{\"id\":\"historical-jun-010\",\"title\":\"爱建证券-电子行业专题报告：Vera Rubin量产提速，RTX Spark打开终端AI新空间-260608.pdf\",\"sourceType\":\"curated_item\",\"time\":\"2026-06\",\"matchedTerms\":[\"PDF\"],\"rank\":-6.203557389986544}]"
AI摘要: "研究者开发并系统评估了针对AMD和DME病变的2D/3D OCT分割流程，在域内验证集上Dice达0.76–0.82。在无病灶标注的外部临床队列OLIVES上，预测虽能跟踪临床生物标志物，但关联弱于域内，提示该工具仍需验证。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-08-30T18:08:37.618Z"
采集批次: "2026年8月30日14点11分37秒"
采集批次ID: "20260830-141137-130"
去重键: "https://arxiv.org/abs/2608.27095"
---

## Computer Science > Computer Vision and Pattern Recognition

## Title:Automated 2D and 3D Segmentation of AMD and DME Lesions in OCT

Authors:[Lucia Sundberg](https://arxiv.org/search/cs?searchtype=author&query=Sundberg,+L), [Zhihao Zhao](https://arxiv.org/search/cs?searchtype=author&query=Zhao,+Z), [M. Ali Nasseri](https://arxiv.org/search/cs?searchtype=author&query=Nasseri,+M+A)

[View PDF](https://arxiv.org/pdf/2608.27095) [HTML (experimental)](https://arxiv.org/html/2608.27095v1)

> Abstract:Age-related macular degeneration (AMD) and diabetic macular edema (DME) are leading causes of vision loss, and optical coherence tomography (OCT) is the standard modality for detecting and monitoring the subtle lesions that drive treatment decisions. Most deep-learning segmentation work for OCT is validated only in-domain, leaving generalization to clinical data collected under different acquisition protocols largely untested. This work develops and systematically ablates four lesion-segmentation pipelines -- 2D and 3D variants for AMD and DME -- reaching Dice scores of 0.76 to 0.82 with strong volumetric and surface calibration (r vol, r surf greater than or equal to 0.97 across all four pipelines) on an in-domain validation set. The ablation process establishes a full-volume, calibration-aware adoption standard that catches mechanisms an ordinary slice-level evaluation would keep, and identifies ensemble composition as the most consistent driver of improvement. To test generalization, the models are evaluated on OLIVES, an external clinical cohort with no lesion-level ground truth, using a proxy-metric framework built around biomarker AUROC, central subfield thickness (CST) correlation, and longitudinal concordance. Predictions track clinical biomarkers outside the training distribution, though less strongly than in-domain -- evidence for, not validation of, automated lesion-burden tracking as a clinical tool.

| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| --- | --- |
| Cite as: | [arXiv:2608.27095](https://arxiv.org/abs/2608.27095) \[cs.CV\] |
|  | (or [arXiv:2608.27095v1](https://arxiv.org/abs/2608.27095v1) \[cs.CV\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.27095](https://doi.org/10.48550/arXiv.2608.27095) |

## Submission history

From: Zhihao Zhao \[[view email](https://arxiv.org/show-email/4f377abf/2608.27095)\]  
**\[v1\]** Thu, 27 Aug 2026 13:13:18 UTC (15,326 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.27095) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))