深度学习模型用于评估尿膀超声图像的质量,使用多级和高阶处理
概括
我们开发了USQNet,这是一种用于自主超声波图像质量评估的深度学习模型. USQNet准确评估图像质量,优于现有方法,并帮助超声波仪解读超声波图像.
科学领域:
- 医疗成像医学成像
- 人工智能的人工智能
- 计算机视觉 计算机视觉
背景情况:
- 自主超声波图像质量评估 (US-IQA) 对于临床解释和机器人程序至关重要.
- 美国IQA的挑战包括图像工件 (噪音,探头定位错误) 和患者特定的解剖变异.
研究的目的:
- 开发一个深层卷积神经网络 (CNN),USQNet,用于类似于超声波的自主超声波图像质量评估.
- 通过使用多尺度和局部到全球二级聚合 (MS-L2GSoP) 分类器来解决US-IQA中的挑战.
主要方法:
- USQNet采用了CNN架构,并使用了新的MS-L2GSoP分类器.
- MS-L2GSoP分类器提取了解剖变异的多尺度特征,并使用二级聚合 (SoP) 来捕获统计依赖.
- 验证是在人类泌尿膀超声图像的新数据集上进行的,与放射科医生评估和最先进的CNN进行比较.
主要成果:
- 在自主超声波图像质量评估中,USQNet实现了92.4%的高精度.
- 该模型的性能比现有的最先进的美国-IQA CNNs 优于3% -14%.
- USQNet显示了与其他模型相比的可比计算时间.
结论:
- USQNet为自主超声波图像质量评估提供了强大而准确的解决方案.
- 开发的模型可以帮助超声波仪,并推进超声波程序的机器人化.
- 该MS-L2GSoP分类器有效地捕获图像特征可靠的US-IQA.
相关概念视频
Ultrasonography
4.5K
Ultrasonography is an imaging technique that uses high-frequency sound waves to visualize the body's internal structures. It is a non-invasive and safe procedure that does not involve the use of ionizing radiation, making it widely used in various medical fields. Ultrasonography is used to study heart function, blood flow in the neck or extremities, certain conditions such as gallbladder disease, and fetal growth and development.
During an ultrasonography procedure, a handheld device called...
During an ultrasonography procedure, a handheld device called...
4.5K
Urinary Bladder
655
The urinary bladder is a hollow, muscular sac that temporarily stores urine before it is expelled from the body. It can hold approximately 600 mL of urine prior to micturition. The bladder is retroperitoneal and located behind the pubic symphysis in the pelvic floor.
In males, the bladder is situated in front of the rectum, while in females, it is positioned anterior to the vagina and uterus. The bladder floor contains an inverted triangular area called the trigone, defined by the two ureteric...
In males, the bladder is situated in front of the rectum, while in females, it is positioned anterior to the vagina and uterus. The bladder floor contains an inverted triangular area called the trigone, defined by the two ureteric...
655


