当前位置: X-MOL 学术J. Digit. Imaging › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Lens Identification to Prevent Radiation-Induced Cataracts Using Convolutional Neural Networks.
Journal of Digital Imaging ( IF 4.4 ) Pub Date : 2019-08-01 , DOI: 10.1007/s10278-019-00242-y
Ross Filice 1
Affiliation  

Exposure of the lenses to direct ionizing radiation during computed tomography (CT) examinations predisposes patients to cataract formation and should be avoided when possible. Avoiding such exposure requires positioning and other maneuvers by technologists that can be challenging. Continuous feedback has been shown to sustain quality improvement and can remind and encourage technologists to comply with these methods. Previously, for use cases such as this, cumbersome manual techniques were required for such feedback. Modern deep learning methods utilizing convolutional neural networks (CNNs) can be used to develop models that can detect lenses in CT examinations. These models can then be used to facilitate automatic and continuous feedback to sustain technologist performance for this task, thus contributing to higher quality patient care. This continuous evaluation for quality purposes also surfaces other operational or process-based challenges that can be addressed. Given high-performance characteristics, these models could also be used for other tasks such as population health research.

中文翻译:

使用卷积神经网络识别镜头以防止辐射引起的白内障。

在计算机断层扫描(CT)检查期间,将镜片暴露于直接电离辐射下会使患者容易患白内障,因此应尽可能避免。避免这种暴露需要技术人员进行定位和其他操作,这可能会具有挑战性。持续反馈已被证明可以保持质量的提高,并且可以提醒和鼓励技术人员遵守这些方法。以前,对于这样的用例,这种反馈需要繁琐的手动技术。利用卷积神经网络(CNN)的现代深度学习方法可用于开发可在CT检查中检测晶状体的模型。然后,这些模型可用于促进自动和连续反馈,以维持技术人员对此任务的性能,从而有助于提高患者护理质量。这种出于质量目的的持续评估还面临其他可以解决的基于操作或基于过程的挑战。考虑到高性能的特征,这些模型也可以用于其他任务,例如人口健康研究。
更新日期:2019-11-01
down
wechat
bug