From 2D to 3D: Automated Ultrasound Segmentation and Cross-Sectional Validation in Murine Tumor Models - replication data
Bibliographic record
Abstract
The dataset comprises B-mode ultrasound images of subcutaneous murine tumors acquired in two preclinical models. Human LN229 glioblastoma multiforme cells were inoculated into the interscapular fat pad of female BALB/c AnN-Foxn1nu/nu/Rj mice, and mouse mammary carcinoma 4T1 cells were inoculated into the mammary fat pad of female BALB/cAnNRj mice (Janvier Labs, Le Genest-Saint-Isle, France). Anatomical B-mode images were acquired using a Vevo F2 ultrasound system with a 25–57 MHz transducer and a Vevo 2100 system with an MS-550D transducer (VisualSonics, Toronto, ON, Canada). All procedures were approved by the Second Local Ethics Committee of Cracow (Permission No. 165/2023 and 250/2020). The original material consisted of (i) individual 2D B-mode images with corresponding expert-drawn tumor masks and (ii) ultrasound videos acquired during freehand examinations, with or without frame-wise annotations. From these videos we extracted individual frames and, where available, the corresponding segmentation contours. The training dataset ultimately used for model development contained 565 images from the first (static-image) source and 2,877 frames extracted from videos; among the latter, 986 frames were provided with tumor masks and 1,891 frames had no mask because either no tumor was visible or the presence of a tumor in that frame was deemed too uncertain by the expert. Images were randomly divided into training, validation, and test subsets for quantitative evaluation. To better characterize model robustness and out-of-distribution behavior, we additionally assembled two curated test resources. First, we selected 10 challenging validation images (e.g. low signal-to-noise ratio, strong contrast heterogeneity within the ROI, or very small tumors) and obtained independent manual segmentations from an expert who was not involved in dataset generation and typically works with different tumor types. Second, we created a “special” testing dataset consisting of 258 masked images from two mice that were not included in the training cohort. For each of these animals, B-mode images were acquired at two time points and in both axial and sagittal planes. In one mouse, LN229 tumors were located directly under the skin rather than in the interscapular fat pad, introducing a tumor location and appearance not represented in the training data. Together, these components provide a diversified benchmark for murine B-mode ultrasound segmentation, spanning multiple tumor models, implantation sites, image qualities, and distribution shifts.
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 | 0.004 | 0.007 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.005 | 0.005 |
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".