Image Production and Evaluation
Image Production and Evaluation in CT tests reconstruction, display, quality metrics, and artifact recognition. The goal is to judge whether an image is diagnostic and to explain why a parameter change helps or hurts.
Core rules
- Reconstruction FOV and matrix determine pixel size: pixel = FOV / matrix. A smaller FOV with the same matrix improves displayed detail.
- Kernels shape noise and resolution: soft-tissue kernels are smooth; bone and lung kernels are sharp and noisier. Iterative reconstruction reduces noise but can alter texture.
- Window level sets the center; window width sets the displayed HU range. Windowing changes display only, not raw data.
- Thin slices improve MPR and small-lesion detection but increase noise unless mA or iterative strength is adjusted.
Worked example
An abdominal image is displayed at level 40 HU and width 400 HU. The window spans 40 ± 200 HU, or −160 to 240 HU. Fat near −100 HU is visible, and contrast-enhanced vessels near 150 HU remain bright. If you narrow the width to 200 HU, the same data look noisier because fewer HU values are mapped to the gray scale.
Traps
- Changing window settings cannot recover information lost to motion or photon starvation.
- A sharp kernel does not automatically improve diagnosis; it may amplify noise.
- Ring artifacts suggest detector calibration problems, not patient motion.
- Motion appears as blurring or streaks; metal causes bright/dark streaks.