chn123/btb3d-ctrate-500-reports

🩺 BTB3D: Automated Radiology Report Generation on CT-RATE (500 Series Evaluation)

Official benchmark evaluation and generated report dataset using BTB3D (Better Tokens for Better 3D, NeurIPS 2025) on 500 randomly sampled 3D Chest CT series from the CT-RATE validation set.

• Model Architecture: 3D MAGViT-2 Tokenizer (16x16x8, 3rd stage) + LLaVA-LLaMA-3.1-8B-Instruct (LoRA Checkpoint-38000).
• Reference Dataset: ibrahimhamamci/CT-RATE
• Model Checkpoints: Forithmuscom/BTB3D
• Hardware: 3x NVIDIA A100-SXM4-40GB GPUs in parallel (~68 minutes total runtime).

---

📊 Benchmark NLP Evaluation Metrics (500 Series)

Evaluated against ground-truth human radiology reports (Findings_EN + Impressions_EN):

Evaluation Metric	BTB3D (500 Scans)	CT2Rep Baseline
BLEU-1	33.18% (0.3318)	~18.4%
BLEU-2	22.26% (0.2226)	~11.2%
BLEU-3	15.64% (0.1564)	~7.1%
BLEU-4	11.40% (0.1140)	~4.6%
ROUGE-1 F1	46.65% (0.4665)	~31.2%
ROUGE-1 Precision	51.86% (0.5186)	~34.5%
ROUGE-1 Recall	47.32% (0.4732)	~29.0%
ROUGE-2 F1	22.20% (0.2220)	~11.8%
ROUGE-L F1	28.31% (0.2831)	~21.5%
METEOR	0.2859	~0.192

---

🫁 Clinical Pathology Concordance & Recall (500 Series)

Pathology / Finding	Ground Truth Positive	Generated Positive	Sensitivity (Recall)	Specificity	F1-Score
Lymphadenopathy / Lymph Nodes	477	454	90.78%	8.70%	93.02%
Lung Nodule / Mass / Lesion	486	447	89.09%	0.00%	92.82%
Pleural Effusion	462	443	88.53%	10.53%	90.39%
Ground Glass / Consolidation / Pneumonia	349	340	69.34%	35.10%	70.25%
Calcification / Atheroma	191	113	26.70%	79.94%	33.55%
Atelectasis	124	102	25.00%	81.12%	27.43%
Bronchiectasis	50	47	16.00%	91.33%	16.49%

---

📁 Repository Contents

• generated_reports_500.csv: Full table of 500 cases containing:
  • VolumeName: CT volume filename (e.g. valid_130_a_1.nii.gz).
  • Generated_Report: Full radiology text report generated by BTB3D.
  • Ground_Truth_Findings: Original English radiologist findings.
  • Ground_Truth_Impressions: Original English radiologist impression.
  • Processing_Time_Sec: Inference and tokenization time in seconds.
• evaluation_metrics.json: Exact quantitative scores (BLEU, ROUGE, METEOR).
• pathology_concordance_500.csv: Statistical analysis across 9 key thoracic pathologies.
• sampled_500_validation_reports.csv: Sampling manifest with voxel spacing, slice count, and metadata.
• run_btb3d_500_pipeline.py: Production-grade multi-GPU pipeline code.
• evaluate_reports.py: Metric evaluation script.

---

📚 References & Upstream Work

This repository contains experimental evaluations and structured radiology reports generated from upstream foundation models and public clinical cohorts. These works were developed by their respective authors. If you refer to or use the underlying models and datasets in your research, please credit and cite the original publications:

@misc{hamamci2025btb3d,
  title  = {Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging},
  author = {Ibrahim Ethem Hamamci and Sezgin Er and Suprosanna Shit and Hadrien Reynaud and Dong Yang and Pengfei Guo and Marc Edgar and Daguang Xu and Bernhard Kainz and Bjoern Menze},
  year   = {2025},
  note   = {Preprint}
}

@misc{hamamci2024ctrate,
  title  = {Generalist Foundation Models from a Multimodal Dataset for 3D Computed Tomography}, 
  author = {Ibrahim Ethem Hamamci and Sezgin Er and Chenyu Wang and Furkan Almas and Ayse Gulnihan Simsek and Sevval Nil Esirgun and Irem Dogan and Omer Faruk Durugol and Benjamin Hou and Suprosanna Shit and Weicheng Dai and Murong Xu and Hadrien Reynaud and Muhammed Furkan Dasdelen and Bastian Wittmann and Tamaz Amiranashvili and Enis Simsar and Mehmet Simsar and Emine Bensu Erdemir and Abdullah Alanbay and Anjany Sekuboyina and Berkan Lafci and Ahmet Kaplan and Zhiyong Lu and Malgorzata Polacin and Bernhard Kainz and Christian Bluethgen and Kayhan Batmanghelich and Mehmet Kemal Ozdemir and Bjoern Menze},
  year   = {2024},
  note   = {Preprint}
}