人工智能
图像质量
公制(单位)
模式识别(心理学)
计算机科学
图像(数学)
计算机视觉
相互信息
探测理论
二值图像
灵敏度(控制系统)
信号(编程语言)
数学
观察员(物理)
度量(数据仓库)
图像处理
接收机工作特性
图像复原
二进制数
补语(音乐)
失真(音乐)
熵(时间箭头)
信号处理
数据挖掘
二进制数据
特征检测(计算机视觉)
质量(理念)
信息论
信噪比(成像)
作者
Changjie Lu,Sourya Sengupta,Hua Li,Mark A. Anastasio
标识
DOI:10.1109/tmi.2026.3680092
摘要
Objective, task-based measures of image quality (IQ) have been widely advocated for assessing and optimizing medical imaging technologies. Besides signal detection theory-based measures, information-theoretic quantities have been proposed to quantify task-based IQ. For example, task-specific information (TSI), defined as the mutual information between an image and a task variable, represents an optimal measure of how informative an image is for performing a specified task. However, like the ideal observer from signal detection theory, TSI does not quantify the amount of task-relevant information in an image that can be exploited by a sub-ideal observer. A recently proposed relaxation of TSI, termed predictive $\mathcal {V}$ -information ( $\mathcal {V}$ -info), removes this limitation and can quantify the utility of an image with consideration of a specified family of sub-ideal observers. In this study, for the first time, we introduce and investigate $\mathcal {V}$ -info as an objective, task-specific IQ metric. To corroborate its usefulness, a stylized magnetic resonance image restoration problem is considered in which $\mathcal {V}$ -info is employed to quantify signal detection or discrimination performance. The presented experiments show that, for binary classification tasks, $\mathcal {V}$ -info varies consistently with the area under the receiver operating characteristic (ROC) curve in regimes where class separability changes with observer capacity or imaging conditions. However, unlike AUC, $\mathcal {V}$ -info remains sensitive in regimes where discrimination performance approaches saturation. In addition, $\mathcal {V}$ -info is readily applicable to multi-class ( $\gt {2}$ ) tasks where ROC analysis is less natural. These findings suggest that $\mathcal {V}$ -info can serve as a complementary task-based image quality measure alongside traditional signal detection theory-based metrics.
科研通智能强力驱动
Strongly Powered by AbleSci AI