Brain Tumor Classification Eval

---
name: brain-tumor-classification-eval
description: Evaluates the ability of machine learning models to classify brain MRI images into four categories: glioma, meningioma, pituitary tumor, and no tumor. It tests feature extraction and decision fusion capabilities using both deep learning and traditional ML classifiers. Use when the user wants to benchmark on Kaggle Brain Tumor MRI dataset, Figshare Brain Tumor dataset, or asks about evaluating this task. Reports Accuracy.
metadata:
  skill_kind: dataset_eval
  source_arxiv: 2603.28357
  bibtex_key: vu2026optimized
  confidence: high
---

# brain-tumor-classification-eval

> Optimized Weighted Voting System for Brain Tumor Classification Using MRI Images — Vu et al. (2026) (arXiv:2603.28357, 2026)

## What this evaluates

Evaluates the ability of machine learning models to classify brain MRI images into four categories: glioma, meningioma, pituitary tumor, and no tumor. It tests feature extraction and decision fusion capabilities using both deep learning and traditional ML classifiers.

## Datasets

- **Kaggle Brain Tumor MRI dataset** — total 7023; splits: train (5712), test (1311); repo https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset
- **Figshare Brain Tumor dataset** — total 3064; splits: train (-1), test (-1); repo https://figshare.com/articles/dataset/brain_tumor_dataset/1512427

## Metrics

- `Accuracy` **(primary)** — range: percent
  - Ratio of correctly classified instances to the total number of instances, multiplied by 100.
- `F1-score` — range: [0, 1]
  - Harmonic mean of precision and recall: 2 * (precision * recall) / (precision + recall).

## Input / output format

**Input**: T1-weighted MRI images preprocessed with balance contrast enhancement, K-means clustering, and Canny edge detection.

**Output**: One of four class labels: 'glioma', 'meningioma', 'pituitary', or 'no tumor'.

## Scoring recipe

```python
def compute_accuracy(predictions, gold_labels):
    correct = sum(1 for p, g in zip(predictions, gold_labels) if p == g)
    return (correct / len(gold_labels)) * 100
```

## Common pitfalls

- Figshare dataset has an imbalanced class distribution, which can skew accuracy metrics compared to the balanced Kaggle dataset.
- The evaluation uses a fixed train-test split rather than k-fold cross-validation, potentially affecting generalizability estimates.
- Image preprocessing steps (edge detection, contrast enhancement) are applied before model input but are not standardized across all baseline comparisons.

## Evidence (verbatim from paper)

> Overall, models trained on the Kaggle dataset exhibited higher accuracy than those trained on Figshare, mainly due to Kaggle’s larger dataset size and more balanced class distribution.

## Citation

```bibtex
@misc{vu2026optimized,
  title={Optimized Weighted Voting System for Brain Tumor Classification Using MRI Images},
  author={Vu et al. (2026)},
  year={2026},
  note={arXiv:2603.28357}
}
```

- arXiv: 2603.28357