Ablation Experiment Results for a Hybrid CNN-Transformer Medical Image Segmentation Model

Bo Li published results from ablation experiments for a dropout method on 2026-04 -29. The dataset is 5.5 KB in size and is stored in an XLS file. The experiments were part of a study proposing a novel encoder combining CNN and Swin Transformer for medical image segmentation, evaluated on the Synapse Multi-Organ and Aortic Vessel Tree datasets.

Use Cases

• Analyzing the impact of dropout techniques based on the described ablation experiments.
• Benchmarking model components based on the described performance metrics (Dice coefficient, Hausdorff Distance).
• Reproducing or validating the performance gains of the proposed HybridNet architecture based on the provided experimental results.

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Results are linked to a publicly available code repository on GitHub.
• Experimental results include specific performance metrics (e.g., Dice coefficients of 84.19% and 87.91%).

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

figshare, author Bo Li

Collection Method

Likely contains numerical results from ablation experiments for a deep learning model.

Time Range

2026

Freshness

Last updated 2026-04-29 17:57:22; freshness should be verified.

Geography

null