nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation
PulseAugur coverage of nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation — every cluster mentioning nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation across labs, papers, and developer communities, ranked by signal.
2 天有情绪数据
-
SegGuidedNet 通过注意力监督改进脑肿瘤分割
研究人员开发了 SegGuidedNet，一种新颖的 3D 神经网络，旨在从 MRI 扫描中进行更准确、更具可解释性的脑肿瘤分割。该网络包含一个 SegAttentionGate 模块，可监督子区域注意力图，提高坏死核心、肿瘤周围水肿和增强肿瘤等肿瘤类型之间的可区分性。该方法在基准数据集上取得了高 Dice 分数，优于其他单一模型，并接近集成方法，同时保持轻量级结构以实现临床实用性。
-
研究论文区分了用于人工智能不确定性的交叉验证与深度集成
一篇题为“折叠中的迷失”的新研究论文强调了人工智能研究中关于医学图像分割不确定性估计的一个普遍误解。研究表明，使用K折交叉验证（CV）来形成集成模型，通常被错误地标记为深度集成（DE），这可能导致对不确定性的不准确解读。研究发现，使用相同训练数据但不同随机种子的DE更适合故障检测等可靠性任务，而CV集成模型更适合建模模糊性。
-
AI models struggle with unseen PET/CT tracer combinations despite segmentation gains
The autoPET3 challenge, held in conjunction with MICCAI 2024, focused on automated lesion segmentation in whole-body PET/CT scans, specifically testing compositional generalization. The challenge utilized a large datase…
-
Sparse convolutional networks boost 3D kidney tumor segmentation accuracy and speed
Researchers have developed a novel two-stage 3D segmentation method using submanifold sparse convolutional networks (SSCNs) for more efficient and accurate kidney tumor detection in CT scans. This approach first identif…
-
LM-CartSeg pipeline automates knee MRI cartilage and bone segmentation for radiomics
Researchers have developed LM-CartSeg, an automated pipeline for segmenting knee MRI scans to analyze cartilage and subchondral bone. This system uses two 3D nnU-Net models and geometric rules for accurate compartmental…
-
Primus V2 Transformer architecture sets new state-of-the-art in 3D medical image segmentation
Researchers have developed Primus and PrimusV2, novel Transformer-centric architectures for 3D medical image segmentation that outperform hybrid models. These new architectures address shortcomings in current Transforme…
-
深度学习模型分割CT扫描中的腹膜癌区域
研究人员开发了一种深度学习方法，可从CT扫描中自动分割放射学腹膜癌指数（rPCI）的区域。该研究在62例CT扫描上评估了nnU-Net和Swin UNETR，其中nnU-Net达到了0.82的Dice相似系数，接近人类观察者间的一致性。该方法旨在为评估腹膜转移瘤提供一种非侵入性的、基于影像学的替代方案，以取代目前侵入性的诊断性腹腔镜检查。
-
New CATMIL method improves brain MRI lesion segmentation accuracy
Researchers have developed a new objective function called CATMIL to improve the segmentation of small structures in brain MRI scans. This method combines standard segmentation loss with auxiliary terms that adaptively …