Informatics in radiology: postprocessing pitfalls in using CT for automatic and semiautomatic determination of global...

Area of Science:

• Cardiovascular Imaging
• Medical Physics
• Radiology

Background:

• Computed tomographic (CT) technology has advanced, enabling four-dimensional (4D) imaging of the beating heart.
• These technical improvements include more detector rows, enhanced spatial and temporal resolution, and retrospective gating.
• 4D CT datasets allow for visualization of the heart across different phases, facilitating cardiac function assessment.

Purpose of the Study:

• To evaluate the capabilities of current software packages for automatic or semi-automatic assessment of left ventricular function using 4D CT datasets.
• To understand the range of automation in these software tools, from manual segmentation to fully automatic analysis.

Main Methods:

• Utilized 4D CT datasets of the beating heart.
• Employed various software packages designed for left ventricular function analysis.
• Compared automated, semi-automated, and manual segmentation approaches.

Main Results:

• High reproducibility of functional parameter assessment was reported across studies.
• Low inter-software variability was observed for larger patient cohorts.
• Significant differences in measurements can arise between different software tools analyzing the same dataset.

Conclusions:

• Automated and semi-automated software can effectively assess left ventricular function from 4D CT data.
• Despite high reproducibility and low inter-software variability in large groups, individual results require careful user review.
• Verification of automatically or semi-automatically derived results is crucial for accurate cardiac function assessment.