Relative Location of CT: 384 Features from CT Images for Anatomical Slice Localization

74 patients (43 male, 31 female) contributed 53,500 CT images used to create this dataset. It contains 384 features derived from histograms describing bone structures and air inclusions in CT slices. The numeric class variable represents the relative location of each slice on the body's axial axis, interpolated from manually annotated landmarks.

Use Cases

• Train regression models to predict the anatomical location of a CT slice based on its histogram features.
• Develop algorithms for automated CT volume reconstruction and slice ordering.
• Research feature extraction methods for representing bone and air structures in polar space from medical images.
• Benchmark models for anatomical landmark interpolation and spatial reasoning within 3D medical scans.

Strengths

• Derived from a substantial collection of 53,500 CT images.
• Features are explicitly described as two concatenated histograms (bone and air) in polar space.
• Class variable was constructed with manual landmark annotation and interpolation, suggesting a structured labeling process.
• Dataset is confirmed to have no missing values.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Graf, F., Kriegel, H.-P., Schubert, M., Poelsterl, S., and Cavallaro, A.

Collection Method

Features extracted from CT images; class variable constructed from manual landmark annotation and interpolation.

Time Range

null

Freshness

null

Geography

null