人工智能培训

搜索

论文代码开源:加热SOFTMAX嵌入(Heated-Up Softmax Embedding)

[复制链接]
admin 发表于 2018-9-15 09:49:44 | 显示全部楼层 |阅读模式
admin 2018-9-15 09:49:44 1754 0 显示全部楼层
人工智能论文代码开源:加热SOFTMAX嵌入(Heated-Up Softmax Embedding)请注意该人工智能论文代码开源在github,大部分是python写的,框架可能是tensorflow或者pytorch。全世界每年有超过50万人被诊断患有头颈癌。放射治疗是这种疾病的重要治疗方法,但它需要手动密集描绘危险的放射敏感性(OARs)。这种规划过程可以延迟治疗开始。虽然自动分割算法提供了一种可能省时的解决方案,但定义,量化和实现专家绩效的挑战依然存在。采用深度学习方法,我们展示了一种3D U-Netarchitecture,其性能与专家相似,描绘了大范围的头部和颈部OAR。该模型在常规临床实践中获得的663个鉴定的计算机断层扫描(CT)扫描的数据集上进行训练,并根据共识OAR定义进行分割。我们通过应用于以前未见到该模型的多个国际站点收集的癌症成像档案库提供的24个CT扫描的独立测试集来证明通用性,每个站点由两个独立的专家分组,由21个通常在临床实践中分段的OAR组成。通过适当的验证研究和监管批准,该系统可以提高放疗途径的有效性。
Over half a million individuals are diagnosed with head and neck cancer eachyear worldwide.Radiotherapy is an important curative treatment for thisdisease, but it requires manually intensive delineation of radiosensitiveorgans at risk (OARs).This planning process can delay treatment commencement.While auto-segmentation algorithms offer a potentially time-saving solution,the challenges in defining, quantifying and achieving expert performanceremain.Adopting a deep learning approach, we demonstrate a 3D U-Netarchitecture that achieves performance similar to experts in delineating a widerange of head and neck OARs.The model was trained on a dataset of 663deidentified computed tomography (CT) scans acquired in routine clinicalpractice and segmented according to consensus OAR definitions.We demonstrateits generalisability through application to an independent test set of 24 CTscans available from The Cancer Imaging Archive collected at multipleinternational sites previously unseen to the model, each segmented by twoindependent experts and consisting of 21 OARs commonly segmented in clinicalpractice.With appropriate validation studies and regulatory approvals, thissystem could improve the effectiveness of radiotherapy pathways.论文代码开源:加热SOFTMAX嵌入(Heated-Up Softmax Embedding) OSRM4R21m2Q5xLCK.jpg
URL地址:https://arxiv.org/abs/1809.04430v1     ----pdf下载地址:https://arxiv.org/pdf/1809.04430v1    ----         ----github下载地址:https://github.com/deepmind/tcia-ct-scan-dataset    ----    论文代码开源:加热SOFTMAX嵌入(Heated-Up Softmax Embedding)请注意该人工智能论文代码开源在github,大部分是python写的,框架可能是tensorflow或者pytorch,keras,至于具体是哪一个没有完全测试。
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

admin当前离线
管理员

查看:1754 | 回复:0

快速回复 返回顶部 返回列表