使用组织微阵列进行自动化乳腺癌分子亚型化的层次图像金字塔变压器框架
Baizhou Li1,2, Yuting Zhong3, Zehang Xing3
1Department of Pathology, Second Affiliated Hospital, Zhejiang University School of Medicine, Zhejiang, PR China.
The Journal of pathology
|January 27, 2026
概括
一个新的深度学习框架,病理学乳腺癌层次图像金字塔转换器 (PBC-HIPT),准确地从H&E染色乳腺癌图像中执行分子亚型和生物标志物预测. 这种自动化方法克服了传统方法的局限性,提供了更高的诊断精度.
科学领域:
- 计算病理学计算病理学
- 人工智能在瘤学中的应用
- 数字病理学和深度学习
背景情况:
- 乳腺癌的分子异质性使诊断和治疗复杂化.
- 目前基于免疫组织化学的亚型分类面临由于内异质性和采样偏差的限制.
- 现有的深度学习模型在高分辨率病理图像中与多尺度特征和远程依赖性作斗争.
研究的目的:
- 开发和验证一种新的深度学习框架,PBC-HIPT,用于使用H&E染色图像进行乳腺癌的自动分子亚型化.
- 评估框架在预测包括雌激素受体 (ER),孕激素受体 (PR),人体表皮生长因子受体2 (HER2) 和Ki-67.7在内的关键生物标志物的表现.
- 将PBC-HIPT与已建立的多实例学习方法进行比较.
主要方法:
- 开发了病理学乳腺癌层次图像金字塔变压器 (PBC-HIPT),这是一个基于多层变压器的架构.
- 在252个组织微阵列 (TMA) 和46个全幻灯片图像 (WSI) 的多机构队列上训练并验证了模型.
- 使用五倍交叉验证对三,四,五类分子亚型和生物标志物预测进行性能评估.
主要成果:
- 在PBC-HIPT实现了84.3%的平均精度和0.91AUC三类亚型任务 (光线,HER2丰富,三阴性乳腺癌).
- 在生物标志物预测方面表现优异:ER (91.8%准确率,0.97 AUC),PR (92.0%准确率,0.96 AUC),Ki-67 (73.8%准确率,0.81 AUC) 和HER2 (84.6%准确率,0.85 AUC).
- 在TMA上表现出强大的概括,但在WSI跨模式验证中表现下降.
结论:
- PBC-HIPT模型提供了一个强大的和自动化的解决方案,用于从H&E染色的TMA中准确地确定乳腺癌分子亚型.
- 该框架有效预测关键的预测和预后生物标志物,帮助个性化治疗.
- 需要进一步的研究来改善全幻灯片图像分析的跨模式概括.
相关概念视频
Bacterial Transformation
59.7K
In 1928, bacteriologist Frederick Griffith worked on a vaccine for pneumonia, which is caused by Streptococcus pneumoniae bacteria. Griffith studied two pneumonia strains in mice: one pathogenic and one non-pathogenic. Only the pathogenic strain killed host mice.
Griffith made an unexpected discovery when he killed the pathogenic strain and mixed its remains with the live, non-pathogenic strain. Not only did the mixture kill host mice, but it also contained living pathogenic bacteria that...
Griffith made an unexpected discovery when he killed the pathogenic strain and mixed its remains with the live, non-pathogenic strain. Not only did the mixture kill host mice, but it also contained living pathogenic bacteria that...
59.7K
Predicting Molecular Geometry
45.7K
VSEPR Theory for Determination of Electron Pair Geometries
45.7K
Molecular Shape and Polarity
75.4K
Dipole Moment of a Molecule
75.4K
DNA Microarrays
21.1K
Microarrays are high-throughput and relatively inexpensive assays that can be automated to analyze large quantities of data at a time. They are used in genome-wide studies to compare gene or protein expression under two varied conditions, such as healthy and diseased states. Microarrays consist of glass or silica slides on which probe molecules are covalently attached through surface functionalization. Most commonly, the slides are prepared through the chemisorption of silanes to silica...
21.1K
Adrenergic Receptors: ɑ Subtype
2.9K
Adrenoceptors are classified into α and ꞵ classes based on their potencies to catecholamine agonists. α-adrenoceptors show the following order of catecholamine potency:
Adrenaline ≥ Noradrenaline >> Isoprenaline
α-adrenoceptors are further divided into α1 and α2-adrenoceptors.
α1-Adrenoceptors: These receptors are located postsynaptically on the effector organs and cause constriction of smooth muscle mediated by activation of phospholipase...
Adrenaline ≥ Noradrenaline >> Isoprenaline
α-adrenoceptors are further divided into α1 and α2-adrenoceptors.
α1-Adrenoceptors: These receptors are located postsynaptically on the effector organs and cause constriction of smooth muscle mediated by activation of phospholipase...
2.9K
Adrenergic Receptors: β Subtype
3.7K
β-adrenoceptors have varied sensitivities towards adrenaline, noradrenaline, and isoprenaline. The order of agonist potency is as follows:
Isoprenaline > Adrenaline > Noradrenaline
Neurotransmitter binding to these receptors causes activation of adenylyl cyclase resulting in increased concentrations of cAMP and modulation of calcium ion channels within the cell. They are further classified into β1, β2, and β3 subtypes.
β1-adrenoceptors: β1-adrenoceptors...
Isoprenaline > Adrenaline > Noradrenaline
Neurotransmitter binding to these receptors causes activation of adenylyl cyclase resulting in increased concentrations of cAMP and modulation of calcium ion channels within the cell. They are further classified into β1, β2, and β3 subtypes.
β1-adrenoceptors: β1-adrenoceptors...
3.7K


