Curriculum Learning for Improved Tumor Segmentation in PET Imaging
Bibliographic record
Abstract
In this paper, we considered the effect of nonuniform sampling of the training data i.e. curriculum learning (CL) on the performance of 3D convolutional neural network for tumor segmentation in PET images. We applied two different curriculums for providing the training data to a convolutional neural network (a 3D U-Net with squeeze and excitation normalization). We applied the easy to hard curriculums by (i) bootstrapping scoring and (ii) self-paced scoring functions. We used augmentation of training data to improve the generalization capability of the trained model and focal loss to take into account the rare samples in any curriculum of training. The learning curves showed that the curriculums based on bootstrapping and self-paced functions speed up the learning while reverse ordering (anti-scoring) makes the training slower and degrades the test performance. The learning by random (uniform) curriculum converges slower. The segmentation results on test data showed that an effective CL via bootstrapping improved segmentation performance and outperformed the trained models obtained via SPS and random curriculums (Dice_bootsrtapping=0.78±0.05 vs. Dice_random=0.72±0.17 vs. Dice_sel-paced=0.63 ± 0.1). In fact, SPS based approach showed a lower mean dice score compared to random curriculum. Anti-Scoring curriculum had the lowest performance in terms of Dice score (0.51±0.21), that confirmed the effect of curriculum on the learning performance.
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 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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".