A novel method for developing contrast-detail curves from clinical patient images based on statistical low-contrast detectability.

This study aimed to develop a method to extract statistical low-contrast detectability (LCD) and develop contrast-detail (C-D) curves from clinical patient images.
Method: The LCD measurement and C-D curve development on the patient images were carried out in the region of air surrounding the patient as an alternative for a homogeneous region within the patient. A simple graphical user interface (GUI) was created to set the initial configuration for interest (ROI), ROI size, and minimum-detectable contrast (MDC). The process was started by segmenting with a threshold between -980 HU and -1024 HU to get an air mask. The mask was trimmed from the patient center coordinates to avoid distortion from the table scan. The mask was used to automatically place square ROIs with a predetermined size. The mean pixel values in Hounsfield units (HU) within each ROI was calculated. Next, the standard deviation (SD) from all the means was obtained. The for a particular target size was generated by multiplying SD by 3.29. A C-D curve was obtained by iterating this process for other ROI sizes. The method was applied to the homogeneous phantom to find the correlation of the parameters inside and outside of the phantom, and implemented on 30 patient images.
Results: Phantom images show a very strong correlation between LCDs obtained from outside and inside the phantom, with R2 of 0.97, 0.96, 0.92, 0.93, 0.80, and 0.88 for tube currents of 80, 100, 120, 140, 160, and 200 mA, respectively. This showed that the air region can act as a surrogate for a homogenous region in the phantom to obtain the LCD and C-D curve. 
Conclusion: The C-D curves obtained from outside the ACR phantom show a strong linear correlation with those from inside the phantom.

Informations de copyright

© 2024 IOP Publishing Ltd.