Using Deep Learning to Predict Final HER2 Status in Invasive Breast Cancers That are Equivocal (2+) by Immunohistochemistry
Bibliographic record
Abstract
Invasive breast carcinomas are routinely tested for HER2 using immunohistochemistry (IHC), with reflex in situ hybridization (ISH) for those scored as equivocal (2+). ISH testing is expensive, time-consuming, and not universally available. In this study, we trained a deep learning algorithm to directly predict HER2 gene amplification status from HER2 2+ IHC slides. Data included 115 consecutive cases of invasive breast carcinoma scored as 2+ by IHC that had follow-up HER2 ISH testing. An external validation data set was created from 36 HER2 IHC slides prepared at an outside institution. All internal IHC slides were digitized and divided into training (80%), and test (20%) sets with 5-fold cross-validation. Small patches (256×256 pixels) were randomly extracted and used to train convolutional neural networks with EfficientNet B0 architecture using a transfer learning approach. Predictions for slides in the test set were made on individual patches, and these predictions were aggregated to generate an overall prediction for each slide. This resulted in a receiver operating characteristic area under the curve of 0.83 with an overall accuracy of 79% (sensitivity=0.70, specificity=0.82). Analysis of external validation slides resulted in a receiver operating characteristic area under the curve of 0.79 with an overall accuracy of 81% (sensitivity=0.50, specificity=0.82). Although the sensitivity and specificity are not high enough to negate the need for reflexive ISH testing entirely, this approach may be useful for triaging cases more likely to be HER2 positive and initiating treatment planning in centers where HER2 ISH testing is not readily available.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".