Glioma Classification Features from Brain Histology Patches

Patch-level features extracted for classifying glioma brain tumors from the IPD-Brain dataset. The dataset contains pre-extracted features, though the total number of patches, features, and rows is unspecified. The original data source is the IPD-Brain project, but the author, organization, and last update date are unknown.

Use Cases

• Train a classifier to distinguish glioma subtypes using the pre-extracted patch-level features.
• Analyze feature importance for glioma diagnosis by evaluating the contribution of individual patch features to model predictions.
• Develop a multi-instance learning model for whole-slide image analysis by aggregating predictions from multiple patch features.
• Benchmark feature extraction methods by comparing the performance of these pre-computed features against newly extracted ones for glioma classification.

Strengths

• Features are pre-extracted, reducing computational load for model training.
• Data originates from the IPD-Brain project, a known source for brain tumor histology.

Limitations

• The sample size, number of features per patch, and total row count are unknown.
• Potential class imbalance or label noise is not described.
• The temporal and geographic coverage of the source histology slides is unspecified.

Provenance

Source

IPD-Brain project

Collection Method

Pre-extracted features from histology image patches.