Incorporating Wave-ViT for Breast Cancer Diagnosis Using MRI Imaging
Bibliographic record
Abstract
Breast cancer remains one of the leading causes of mortality among women globally, and early detection is critical for improving survival rates. Breast MRI, the most sensitive imaging modality for detection, often involves manual review of numerous slices, which is time-intensive and prone to human error. Machine learning (ML) algorithms offer a transformative solution by automating this process, improving efficiency, and enhancing diagnostic accuracy. In this study, we propose a machine learning approach to enhance breast cancer prediction and diagnosis. We utilize a pre-trained multiscale vision transformer, Wave-ViT, to classify MRI slices as healthy or unhealthy. The model was trained and tested on MRI scans from 922 patients in the Duke Breast Cancer MRI dataset and independently validated on 143 patients from the MAMA-MIA dataset. To ensure high-quality data, both datasets were carefully curated to exclude noisy or mislabeled slices. The model's performance was evaluated using accuracy, F1-score, precision, recall, and confusion matrices under various experimental conditions. These included randomized training and testing splits using the Fisher-Yates shuffle, exploration of different Wave-ViT variants, and testing across multiple training set configurations. Our approach consistently demonstrated over 94\% accuracy on the external validation dataset, showcasing the potential of machine learning algorithms like Wave-ViT to reduce diagnostic workloads and improve breast cancer detection outcomes.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".