眼动追踪技术在孤独症谱系障碍的早期诊断中具有潜在的应用价值.为研究孤独症儿童对不同面孔加工的特点,应用机器学习算法对其进行自动识别,本研究选取3~6岁孤独症儿童40名和性别、年龄相匹配的正常儿童41名观看异国陌生面孔、本国陌生...眼动追踪技术在孤独症谱系障碍的早期诊断中具有潜在的应用价值.为研究孤独症儿童对不同面孔加工的特点,应用机器学习算法对其进行自动识别,本研究选取3~6岁孤独症儿童40名和性别、年龄相匹配的正常儿童41名观看异国陌生面孔、本国陌生面孔和本国熟悉面孔,根据两组儿童眼动坐标数据,使用机器学习算法进行自动划分兴趣区、特征选择和分类,来判断不同面孔的扫描模式是否可以用于识别孤独症儿童,并从准确率、特异性、敏感性和可靠性4个方面对分类模型进行评估.结果显示,基于不同面孔扫描模式的机器学习算法可以提取足够的信息来区分孤独症和正常儿童,最大分类准确率为90.28%,对应AUC(area under the ROC curve)为0.9317.因此,眼动追踪技术结合机器学习能够为临床诊断提供辅助的评价指标.展开更多
Single-cell sequencing has been constrained by the trade-off among throughput,capture bias,and compatibility with cells of unusual size or morphology.A recent innovative approach,Stereo-cell,addresses these constraint...Single-cell sequencing has been constrained by the trade-off among throughput,capture bias,and compatibility with cells of unusual size or morphology.A recent innovative approach,Stereo-cell,addresses these constraints by coupling high-density DNA nanoball patterned arrays with planar in situ RNA capture and microscopy-guided segmentation,thereby eliminating droplet encapsulation while scaling the field of view by chip size.This planar architecture scales from thousands to>106 cells per chip,maintains robust RNA in situ capture,and natively integrates multiplex immunofluorescence and oligo-barcoded antibodies to deliver concurrent transcriptomic and proteomic readouts.This perspective evaluates Stereo-cell relative to droplet-and plate-based methods across throughput,sensitivity,spatial resolution,and sample versatility,and outlines practical considerations for rarecell detection and subcellular transcript localization.By bridging single-cell and spatial omics in a unified workflow,Stereo-cell offers a general-purpose platform to map cellular states,interactions,and subcellular organization at unprecedented scale.展开更多
文摘眼动追踪技术在孤独症谱系障碍的早期诊断中具有潜在的应用价值.为研究孤独症儿童对不同面孔加工的特点,应用机器学习算法对其进行自动识别,本研究选取3~6岁孤独症儿童40名和性别、年龄相匹配的正常儿童41名观看异国陌生面孔、本国陌生面孔和本国熟悉面孔,根据两组儿童眼动坐标数据,使用机器学习算法进行自动划分兴趣区、特征选择和分类,来判断不同面孔的扫描模式是否可以用于识别孤独症儿童,并从准确率、特异性、敏感性和可靠性4个方面对分类模型进行评估.结果显示,基于不同面孔扫描模式的机器学习算法可以提取足够的信息来区分孤独症和正常儿童,最大分类准确率为90.28%,对应AUC(area under the ROC curve)为0.9317.因此,眼动追踪技术结合机器学习能够为临床诊断提供辅助的评价指标.
基金National Key R&D Plan of China,Grant/Award Number:2023YFB3210400Major Scientific and Technological Innovation Project of Shandong Province,Grant/Award Numbers:2024ZLGX01,2022CXGC020501Shandong University Integrated Research and Cultivation Project,Grant/Award Number:2022JC001。
文摘Single-cell sequencing has been constrained by the trade-off among throughput,capture bias,and compatibility with cells of unusual size or morphology.A recent innovative approach,Stereo-cell,addresses these constraints by coupling high-density DNA nanoball patterned arrays with planar in situ RNA capture and microscopy-guided segmentation,thereby eliminating droplet encapsulation while scaling the field of view by chip size.This planar architecture scales from thousands to>106 cells per chip,maintains robust RNA in situ capture,and natively integrates multiplex immunofluorescence and oligo-barcoded antibodies to deliver concurrent transcriptomic and proteomic readouts.This perspective evaluates Stereo-cell relative to droplet-and plate-based methods across throughput,sensitivity,spatial resolution,and sample versatility,and outlines practical considerations for rarecell detection and subcellular transcript localization.By bridging single-cell and spatial omics in a unified workflow,Stereo-cell offers a general-purpose platform to map cellular states,interactions,and subcellular organization at unprecedented scale.
