Ablation Study Results for CT Kidney Lesion Segmentation Model

5.5 KB of tabular results from an ablation study for the GDT-SwinKid model, a hybrid transformer designed for renal lesion analysis. The dataset, authored by Thirupathi Rao N and last updated in May 2026, contains performance metrics from model validation on clinical CT data. The study reports Dice coefficients as high as 0.95 and AUC values approaching 0.99.

Use Cases

• Comparing segmentation model performance based on reported Dice coefficients and AUC-ROC values.
• Analyzing the impact of architectural components like Gamma distribution-based feature refinement mentioned in the description.
• Benchmarking new kidney lesion analysis methods against the reported 5–9% improvement over U-Net and Swin Transformer baselines.

Strengths

• Specific performance metrics are provided, including Dice coefficients up to 0.95 and AUC-ROC values approaching 0.99.
• The dataset is associated with a model validated on a large set of clinical datasets, as stated in the description.
• It has a clear open license (CC-BY-4.0).

Limitations

• The dataset is very small (5.5 KB), suggesting limited scope or a summary of results rather than raw data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for statistical analysis.

Provenance

Source

Thirupathi Rao N, via figshare.

Collection Method

Likely contains aggregated results from model training and validation experiments.

Freshness

Last updated 2026-05-21 07:33:36; freshness should be verified.