---
license: cc-by-nc-sa-4.0
language:
- en
base_model:
- Qwen/Qwen3-Embedding-0.6B
datasets:
- chest2vec/chest2vec_labels
pipeline_tag: text-classification
library_name: transformers
tags:
- radiology
- chest-ct
- report-labeling
- multi-label
- ct-rate
- chexbert-style-f1
---

# chest2vec CT Report Labeler (0.6B)

A weakly-supervised **multi-label classifier** that reads a free-text **chest-CT report** and
predicts a **137-leaf chest-imaging taxonomy**, with a **ternary** status per label
(*negative / uncertain / positive*).

It also provides a **CheXbert / SRR-BERT-style report-comparison F1**: label a list of
ground-truth reports and a list of generated/predicted reports, then score them against each
other (micro / macro / weighted F1) — useful for evaluating radiology report generation.

- **Base architecture:** [`Qwen/Qwen3-Embedding-0.6B`](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) (Apache-2.0)
- **Adaptation:** LoRA (r=16, α=32) **merged into the weights** + last-token (EOS) pooling + L2-norm + a linear ternary head (`1024 → 137 × 3`)
- **Self-contained:** the full model (encoder + head) ships in `model.safetensors`. Loading does **not** download Qwen3-Embedding weights — the architecture is rebuilt from the bundled config and our weights are loaded in. Tokenizer is bundled too.
- **Params:** ~596M · weights in float32
- **Training labels:** [`chest2vec/chest2vec_labels`](https://huggingface.co/datasets/chest2vec/chest2vec_labels) (revised CT-RATE, 137-leaf taxonomy)

## Label space

137 leaf labels over 9 chest-CT sections (Lungs & Airways, Pleura, Mediastinum & Hila,
Cardiovascular, Chest Wall, Bones/Spine, Upper Abdomen, Lower Neck, Others). The exact list is
in `config.json` (`labels`); full definitions and per-split counts are in the dataset's
[`LABEL_HIERARCHY.md`](https://huggingface.co/datasets/chest2vec/chest2vec_labels/blob/main/LABEL_HIERARCHY.md).

**Ternary head** — `softmax(logits, dim=-1)` over class indices `[0, 1, 2]`:

| class index | meaning | value |
|---:|---|---:|
| 0 | negative | 0 |
| 1 | uncertain | -1 |
| 2 | positive | 1 |

A label is reported **positive** when `P(class=2) ≥ threshold` (default **0.5**).

## Usage

```python
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("chest2vec/chest2vec_labeler", trust_remote_code=True).eval()
tok   = AutoTokenizer.from_pretrained("chest2vec/chest2vec_labeler", trust_remote_code=True)

reports = ["Bibasilar atelectasis with small bilateral pleural effusions. Cardiomegaly. Coronary artery calcification."]

# 1) human-readable positive labels per report
print(model.label_reports(reports, tokenizer=tok))
# [{'Subsegmental / linear atelectasis': 'positive', 'Pleural effusion': 'positive',
#   'Cardiomegaly': 'positive', 'Coronary artery calcification': 'positive'}]

# 2) full prediction matrices
out = model.predict(reports, tokenizer=tok, threshold=0.5, return_ternary=True)
out["labels"]      # list of 137 label names
out["proba"]       # [N, 137] P(positive)
out["positive"]    # [N, 137] in {0,1}
out["ternary"]     # [N, 137] in {-1,0,1}
```

### CheXbert / SRR-BERT-style report comparison

Label both ground-truth and predicted reports, then compute label-level F1 (GT-labels treated
as truth):

```python
res = model.score_reports(gt_reports, pred_reports, tokenizer=tok)   # equal-length lists
print(res["micro"]["f1"], res["macro"]["f1"], res["weighted"]["f1"])
print(res["per_label"]["Pleural effusion"])   # {'precision':..,'recall':..,'f1':..,'support_gt':..}

# or one-liner that loads the model for you:
from modeling_chest2vec_labeler import report_f1
report_f1(gt_reports, pred_reports, tokenizer=tok)
```

Returns `micro`, `macro`, `weighted` precision/recall/F1 over the 137 labels, plus `per_label`.

## Inputs & conventions

- Input is the **findings** text (the model was trained on CT-RATE findings + their refined
  section-structured form). Reports are formatted internally as
  `Instruct: Given the following chest CT report, extract the presence/absence of entities\nQuery: <report>`,
  truncated to **512** tokens, with an EOS token appended and left-padding.
- For best fidelity, run in float32 (default). bf16 is fine for throughput with negligible drift.

## Evaluation

Direct-paragraph evaluation (`softmax positive-class`, macro-F1 over labels with ≥30 positives —
the stable headline; the all-labels macro is dragged down by sparse-tail labels):

| Eval set | reports | macro-F1 (≥30) @0.33 | macro-F1 (≥30) @0.5 | macro-AUC (≥30) |
|---|--:|--:|--:|--:|
| CT-RATE revised test (public) | 1,464 | **0.875** | 0.866 | 0.989 |
| sample1000 (private radiologist-reviewed gold) | 1,000 | **0.766** | 0.761 | 0.972 |

On the public test set, a radiologist reviewed **966 reports**: **857 fully accepted, 60
imperfect-but-acceptable, 49 failed** (94.9% acceptable). AUC barely moves public→private
(0.989 → 0.972), i.e. label ranking transfers to real clinical reports; the F1 gap is mostly
threshold/labeling-convention, not domain failure.

## Caveats

- **Weakly supervised** — trained on LLM-generated labels (not radiologist ground truth) derived
  from report **text**, not images. Not a medical device; not for clinical use.
- `IVC filter` is in the taxonomy for completeness but had no training positives.
- `score_reports` measures **label agreement** between two reports as judged by this labeler;
  like CheXbert-F1 it inherits the labeler's own error modes.

## License & attribution

Released under **CC-BY-NC-SA-4.0**. Built on **`Qwen/Qwen3-Embedding-0.6B`** (Apache-2.0) and
trained using labels derived from **[CT-RATE](https://huggingface.co/datasets/ibrahimhamamci/CT-RATE)**
(CC-BY-NC-SA-4.0). **If you use this model, cite the CT-RATE paper** (arXiv:2403.17834) and
acknowledge Qwen3-Embedding. See the [dataset card](https://huggingface.co/datasets/chest2vec/chest2vec_labels)
for the full citation.