Background: Mesh expansion and fixation at retro-rectus plane through multiples stabs produces good results. But these stabs cause cosmetic disorders for the patients and doctors. So, we find some modification to do t...Background: Mesh expansion and fixation at retro-rectus plane through multiples stabs produces good results. But these stabs cause cosmetic disorders for the patients and doctors. So, we find some modification to do this procedure without these stabbing wounds in midline hernial repair. Patients and methods: This technique was used to fix the mesh at retro-rectus plane in 50 patients suffering from midline hernias, from January 2008 through January 2010 at Zagazig university Hospital, Egypt. Laparotomy incision was done over the hernial sac or at old incision;the contents were then released and reduced into peritoneal cavity without much subcutaneous dissection. The suitable sheet of polypropylene mesh to cover the hernial defect and any weak area was prepared and fixed at retro-rectus plane percutaneously without stabbing wounds by using redirecting suture hook. The mean period of follow up was 26 months. Results: There was no recurrence during the period of follow up. Five patients developed subcutaneous bluish discoloration at the site of some stitches, which disappear within two weeks with conservative treatment. Conclusion: Percutaneous mesh expansion and fixation at retro-rectus plane by using redirecting suture hook procedure has good results in recurrence rate and cosmetic appearance.展开更多
从三维Mesh数据中分割建筑物立面以识别对象,是三维场景理解的关键,但现有方法多依赖高成本的精细标注数据。针对该问题,提出了一种半监督学习方法,引入一种基于对比学习和一致性正则化的半监督语义分割(semi-supervised semantic segme...从三维Mesh数据中分割建筑物立面以识别对象,是三维场景理解的关键,但现有方法多依赖高成本的精细标注数据。针对该问题,提出了一种半监督学习方法,引入一种基于对比学习和一致性正则化的半监督语义分割(semi-supervised semantic segmentation based on contrastive learning and consistency regularization,SS_CC)方法,用于分割三维Mesh数据的建筑物立面。在SS_CC方法中,改进后的对比学习模块利用正负样本之间的类可分性,能够更有效地利用类特征信息;提出的基于特征空间的一致性正则化损失函数,从挖掘全局特征的角度增强了对所提取建筑物立面特征的鉴别力。实验结果表明,所提出的SS_CC方法在F1分数、mIoU指标上优于当前一些主流方法,且在建筑物的墙面和窗户上的分割效果相对更好。展开更多
文摘Background: Mesh expansion and fixation at retro-rectus plane through multiples stabs produces good results. But these stabs cause cosmetic disorders for the patients and doctors. So, we find some modification to do this procedure without these stabbing wounds in midline hernial repair. Patients and methods: This technique was used to fix the mesh at retro-rectus plane in 50 patients suffering from midline hernias, from January 2008 through January 2010 at Zagazig university Hospital, Egypt. Laparotomy incision was done over the hernial sac or at old incision;the contents were then released and reduced into peritoneal cavity without much subcutaneous dissection. The suitable sheet of polypropylene mesh to cover the hernial defect and any weak area was prepared and fixed at retro-rectus plane percutaneously without stabbing wounds by using redirecting suture hook. The mean period of follow up was 26 months. Results: There was no recurrence during the period of follow up. Five patients developed subcutaneous bluish discoloration at the site of some stitches, which disappear within two weeks with conservative treatment. Conclusion: Percutaneous mesh expansion and fixation at retro-rectus plane by using redirecting suture hook procedure has good results in recurrence rate and cosmetic appearance.
文摘从三维Mesh数据中分割建筑物立面以识别对象,是三维场景理解的关键,但现有方法多依赖高成本的精细标注数据。针对该问题,提出了一种半监督学习方法,引入一种基于对比学习和一致性正则化的半监督语义分割(semi-supervised semantic segmentation based on contrastive learning and consistency regularization,SS_CC)方法,用于分割三维Mesh数据的建筑物立面。在SS_CC方法中,改进后的对比学习模块利用正负样本之间的类可分性,能够更有效地利用类特征信息;提出的基于特征空间的一致性正则化损失函数,从挖掘全局特征的角度增强了对所提取建筑物立面特征的鉴别力。实验结果表明,所提出的SS_CC方法在F1分数、mIoU指标上优于当前一些主流方法,且在建筑物的墙面和窗户上的分割效果相对更好。
