目的综合考虑B型超声(B-mode ultrasound,B-US)和对比增强超声(contrast-enhanced ultrasound,CEUS)双模态信息有助于提升乳腺肿瘤诊断的准确性,从而利于提高患者生存率。然而,目前大多数模型只关注B-US的特征提取,忽视了CEUS特征的学...目的综合考虑B型超声(B-mode ultrasound,B-US)和对比增强超声(contrast-enhanced ultrasound,CEUS)双模态信息有助于提升乳腺肿瘤诊断的准确性,从而利于提高患者生存率。然而,目前大多数模型只关注B-US的特征提取,忽视了CEUS特征的学习和双模态信息的融合处理。为解决上述问题,提出了一个融合时空特征与时间约束的双模态乳腺肿瘤诊断模型(spatio-temporal feature and temporal-constrained model,STFTCM)。方法首先,基于双模态信息的数据特点,采用异构双分支网络学习B-US和CEUS包含的时空特征。然后,设计时间注意力损失函数引导CEUS分支关注造影剂流入病灶区的时间窗口,从该窗口期内提取CEUS特征。最后,借助特征融合模块实现双分支网络之间的横向连接,通过将B-US特征作为CEUS分支补充信息的方式,完成双模态特征融合。结果在收集到的数据集上进行对比实验,STFTCM预测的正确率、敏感性、宏平均F1和AUC(area under the curve)指标均表现优秀,其中预测正确率达88.2%,领先于其他先进模型。消融实验中,时间注意力约束将模型预测正确率提升5.8%,特征融合使得模型诊断正确率相较于单分支模型至少提升2.9%。结论本文提出的STFTCM能有效地提取并融合处理B-US和CEUS双模态信息,给出准确的诊断结果。同时,时间注意力约束和特征融合模块可以显著地提升模型性能。展开更多
Objective： Chronic lymphocytic leukemia （CLL） and mantle cell lymphoma （MCL） cells over-express a guanine exchange factor （GEF）, Rasgrf-1. This GEF increases active Ras as it catalyzes the removal of GDP from R...Objective： Chronic lymphocytic leukemia （CLL） and mantle cell lymphoma （MCL） cells over-express a guanine exchange factor （GEF）, Rasgrf-1. This GEF increases active Ras as it catalyzes the removal of GDP from Ras so that GTP can bind and activate Ras. This study aims to study the mechanism of action of Rasgrf-1 in B-cell malignancies. Methods： N-terminus truncated Rasgrf-1 variants have a higher GEF activity as compared to the full-length transcript therefore a MCL cell line with stable over-expression of truncated Rasgrf-1 was established. The B-cell receptor （BCR） and chemokine signaling pathways were compared in the Rasgrf-I over-expressing and a control transfected cell line. Results： Cells over-expressing truncated form of Rasgrf-1 have a higher proliferative rate as compared to control transfected cells. BCR was activated by lower concentrations of anti-IgM antibody in Rasgrf-1 over-expressing cells as compared to control cells indicating that these cells are more sensitive to BCR signaling. BCR signaling also phosphorylates Rasgrf-1 that further increases its GEF function and amplifies BCR signaling. This activation of Rasgrf-1 in over-expressing cells resulted in a higher expression of phospho-ERK, AKT, BTK and PKC-alpha as compared to control cells. Besides BCR, Rasgrf-1 over-expressing cells were also more sensitive to microenvironment stimuli as determined by resistance to apoptosis, chemotaxis and ERK pathway activation. Conclusions： This GEF protein sensitizes B-cells to BCR and chemokine mediated signaling and also upregulates a number of other signaling pathways which promotes growth and survival of these cells.展开更多
文摘目的综合考虑B型超声(B-mode ultrasound,B-US)和对比增强超声(contrast-enhanced ultrasound,CEUS)双模态信息有助于提升乳腺肿瘤诊断的准确性,从而利于提高患者生存率。然而,目前大多数模型只关注B-US的特征提取,忽视了CEUS特征的学习和双模态信息的融合处理。为解决上述问题,提出了一个融合时空特征与时间约束的双模态乳腺肿瘤诊断模型(spatio-temporal feature and temporal-constrained model,STFTCM)。方法首先,基于双模态信息的数据特点,采用异构双分支网络学习B-US和CEUS包含的时空特征。然后,设计时间注意力损失函数引导CEUS分支关注造影剂流入病灶区的时间窗口,从该窗口期内提取CEUS特征。最后,借助特征融合模块实现双分支网络之间的横向连接,通过将B-US特征作为CEUS分支补充信息的方式,完成双模态特征融合。结果在收集到的数据集上进行对比实验,STFTCM预测的正确率、敏感性、宏平均F1和AUC(area under the curve)指标均表现优秀,其中预测正确率达88.2%,领先于其他先进模型。消融实验中,时间注意力约束将模型预测正确率提升5.8%,特征融合使得模型诊断正确率相较于单分支模型至少提升2.9%。结论本文提出的STFTCM能有效地提取并融合处理B-US和CEUS双模态信息,给出准确的诊断结果。同时,时间注意力约束和特征融合模块可以显著地提升模型性能。
文摘Objective： Chronic lymphocytic leukemia （CLL） and mantle cell lymphoma （MCL） cells over-express a guanine exchange factor （GEF）, Rasgrf-1. This GEF increases active Ras as it catalyzes the removal of GDP from Ras so that GTP can bind and activate Ras. This study aims to study the mechanism of action of Rasgrf-1 in B-cell malignancies. Methods： N-terminus truncated Rasgrf-1 variants have a higher GEF activity as compared to the full-length transcript therefore a MCL cell line with stable over-expression of truncated Rasgrf-1 was established. The B-cell receptor （BCR） and chemokine signaling pathways were compared in the Rasgrf-I over-expressing and a control transfected cell line. Results： Cells over-expressing truncated form of Rasgrf-1 have a higher proliferative rate as compared to control transfected cells. BCR was activated by lower concentrations of anti-IgM antibody in Rasgrf-1 over-expressing cells as compared to control cells indicating that these cells are more sensitive to BCR signaling. BCR signaling also phosphorylates Rasgrf-1 that further increases its GEF function and amplifies BCR signaling. This activation of Rasgrf-1 in over-expressing cells resulted in a higher expression of phospho-ERK, AKT, BTK and PKC-alpha as compared to control cells. Besides BCR, Rasgrf-1 over-expressing cells were also more sensitive to microenvironment stimuli as determined by resistance to apoptosis, chemotaxis and ERK pathway activation. Conclusions： This GEF protein sensitizes B-cells to BCR and chemokine mediated signaling and also upregulates a number of other signaling pathways which promotes growth and survival of these cells.
