Skip to content

D Glossary

title: D Glossary
description: Introduces common terminology

English TermExplanation
k-spaceIn MRI, refers to the original frequency domain data matrix collected, through which images are generated by Fourier transform.
T1-weightedA contrast method of MRI sequences, emphasizing longitudinal relaxation characteristics, with clear structure.
T2-weightedAnother contrast method of MRI sequences, emphasizing transverse relaxation characteristics, with obvious display of liquid/edema.
Windowing (Window Width/Level)In CT or X-ray image display, used to adjust grayscale range and midpoint to optimize contrast.
Flat-field correctionCorrects dark spots or streak artifacts in images caused by detector/illumination non-uniformity or inconsistent response.
SENSE (Sensitivity Encoding)MRI parallel imaging technique that uses multiple coil sensitivity information to achieve scanning acceleration.
Parallel ImagingUses multiple receiving coils to sample simultaneously to reduce scanning time or improve resolution.
Echo Time (TE)The time interval between RF pulse and signal acquisition start in MRI, affecting image contrast.
Repetition Time (TR)The pulse cycle interval time in MRI, affecting signal recovery and contrast.
Flip AngleThe tilt angle of nuclear magnetic resonance caused by MRI radio frequency pulse, affecting signal intensity/contrast.
Field of View (FOV)The anatomical region or volume size covered by imaging, usually expressed in mm.
Voxel3D pixel, the smallest unit in medical imaging data.
Slice ThicknessThe thickness of each image in tomographic scanning (CT/MRI), affecting volume reconstruction and partial volume effect.
Partial Volume EffectWhen a voxel contains multiple tissue types, signal mixing leads to reduced accuracy.
ArtifactNon-real tissue structures in images introduced by equipment, patient movement, signal processing, etc.
Signal-to-Noise Ratio (SNR)The ratio of signal intensity to noise intensity in an image, measuring image quality.
Contrast-to-Noise Ratio (CNR)The ratio of signal difference between lesion and background tissue to noise, reflecting recognizability.
Reconstruction KernelThe filter type used during CT reconstruction, which affects sharpness and noise level.
RadiomicsExtracting large amounts of quantitative features (texture, shape, intensity, etc.) from images for modeling and analysis.
SegmentationClassifying image pixels/voxels into different structures (such as organs, edema, tumors) for quantitative analysis.
RegistrationAligning images from different time points or different modalities to the same space for convenient comparison.
Deep LearningUsing deep neural networks (such as CNN, Transformer) to analyze/predict image data.
Transfer LearningApplying a model trained on one task to another related task to reduce training costs.
Multimodal LearningFusing multiple imaging modalities (such as MRI + PET + genetic data) for joint analysis.
OverfittingThe phenomenon where a model performs well on training set but has poor generalization ability on new data.
Dice CoefficientA metric that measures the degree of overlap between predicted and real regions in segmentation tasks.
Intersection over Union (IoU)Used for segmentation evaluation: the ratio of the intersection of prediction and real regions to their union.
Sensitivity (Recall)The proportion of true positives correctly identified in classification tasks.
SpecificityThe proportion of true negatives correctly identified.
ROC CurveA curve showing the relationship between false positive rate and true positive rate, used for classification evaluation.
Hounsfield Unit (HU)Unit representing tissue density in CT: water is 0, air is approximately -1000.
Contrast AgentInjected or taken before imaging to enhance contrast of certain tissues or blood vessels.
Dual-energy CTUsing two different energies of X-rays to contrast and enhance imaging material discrimination capability.
Ground-glass Opacity (GGO)Mildly increased opacity in lung images where vascular texture is still visible.
MassLarger volume (usually >3 cm) possibly abnormal structure or tumor, requiring observation and differentiation.
NoduleSmaller volume (usually ❤️ cm) structure, requiring follow-up for growth or malignancy.
Pleural EffusionFluid accumulation in the pleural cavity, appearing as fluid level or blurred boundary in images.
CardiomegalyIncreased cardiothoracic ratio, suggesting cardiac structural abnormalities.
CalcificationCalcium deposition in tissues or lesions, visible as high density in CT or X-ray.
InfarctTissue necrosis caused by interrupted blood supply, with typical imaging manifestations in MRI/CT.
Ring EnhancementIn contrast-enhanced images, lesion edge enhancement with low signal/low density center, common in metastatic tumors or brain abscesses.
Spiculated MarginTumor edge with radiating spiky protrusions, suggesting invasiveness or malignancy.
GhostingParticularly in the phase encoding direction in MRI images, streaks or duplicate images caused by patient movement, etc.
Flip AngleThe tilt angle of nuclear magnetic resonance caused by MRI radio frequency pulse, affecting signal intensity/contrast.
Fourier TransformMathematical transformation that converts spatial domain signals to frequency domain (such as k-space to image domain).
Coil SensitivityResponse differences of multi-channel MRI receiving coils at different spatial positions, used for parallel imaging.
Nyquist FrequencyHalf of the lowest necessary sampling frequency in sampling theory, below which aliasing will occur.
AliasingWhen sampling is insufficient or FOV is too small, signals outside the image are repeated or overlapped in the image.

Released under the MIT License.