Small Lesions, Big Impact: An Automated Segmentation Framework for Brain Metastases<sup>*</sup>
Bibliographic record
Abstract
Accurate delineation of brain metastases on MRI is crucial for effective radiation treatment planning and outcome evaluation. However, manual segmentation of these images is resource-intensive and time-consuming, especially for patients with multiple brain metastases. Recently, several deep learning-based approaches have been introduced for the segmentation of brain tumors; however, these approaches generally fail to delineate small lesions accurately. In this paper, we have introduced a deep learning framework for automatic segmentation of brain metastases with a focus on improving the accuracy for smaller lesions. Our approach utilizes the strengths of Swin Transformer's attention mechanism with a hierarchical encoder to capture the intricate characteristics of brain metastases on multi-modal MRI precisely. In the context of stereotactic radiotherapy, this precision-focused strategy holds particular significance, where accurate segmentation is a necessity even for tiny tumors. The framework was trained and validated on the BraTS-METS dataset, where it achieved a Dice score and volume estimation error of 84.4 ± 3.4 and 0.37 ± 0.4 cc, respectively, on an independent test set. The performance of the framework was compared with two state-of-the-art segmentation models over different tumor size categories, where it demonstrated a considerable improvement in all experiments. Beyond its technical accomplishments, this study is a step forward towards automating stereotactic radiotherapy planning and outcome evaluation for brain tumors.
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 | 0.001 |
| 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) | 0.002 | 0.002 |
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".