CheXVision-ResNet

CheXVision — Deep Learning & Big Data university project. 14-class chest X-ray pathology detection + binary normal/abnormal classification on the NIH Chest X-ray14 dataset (112,120 images).

Project Resources

Architecture

SE-ResNet architecture

Training Pipeline

Training pipeline

Training Metrics

Best validation macro AUC-ROC: 0.8141
Best validation binary AUC-ROC: 0.7739
Best validation binary F1: 0.6587
Best checkpoint epoch: 41

Per-Class AUC-ROC at Best Epoch

Pathology	AUC-ROC	Visual
Atelectasis	`0.8022`	`████████░░`
Cardiomegaly	`0.9059`	`█████████░`
Effusion	`0.8831`	`█████████░`
Infiltration	`0.7060`	`███████░░░`
Mass	`0.8596`	`█████████░`
Nodule	`0.7525`	`████████░░`
Pneumonia	`0.7298`	`███████░░░`
Pneumothorax	`0.8329`	`████████░░`
Consolidation	`0.8080`	`████████░░`
Edema	`0.9122`	`█████████░`
Emphysema	`0.8545`	`█████████░`
Fibrosis	`0.7622`	`████████░░`
Pleural_Thickening	`0.7782`	`████████░░`
Hernia	`0.8101`	`████████░░`

Training Configuration

Repository: arudaev/chexvision-scratch
Dataset: arudaev/chest-xray-14-320 · revision 44443e6ee968b3c6094b63f14a27698c40b50680
Architecture: Custom residual CNN with Squeeze-Excitation channel attention (depth [3, 4, 6, 3]) trained from scratch with shared features and dual classification heads.
Platform: Kaggle GPU kernel (NVIDIA T4 / P100)
Batch size: 24 × grad_accum 4 = effective batch 96
AMP (fp16): enabled
CLAHE preprocessing: enabled
Label smoothing: 0.1
Optimizer: AdamW · Scheduler: CosineAnnealingLR
Epochs configured: 100 · Early stop patience: 15

Intended Use

This model is intended for research and educational work on automated chest X-ray pathology detection. It outputs two predictions per image:

Multi-label scores — independent sigmoid probability for each of 14 NIH pathologies
Binary score — sigmoid probability of any abnormality (Normal vs. Abnormal)

Limitations

Not validated for clinical use. Predictions must not substitute professional medical judgment.
Trained on NIH Chest X-ray14, which contains noisy radiologist annotations (patient-level labels, not lesion-level).
Performance degrades on images from equipment, patient populations, or preprocessing pipelines that differ from the NIH training distribution.
Reported AUC metrics are on the validation split, not the held-out test set.

CheXNet Benchmark Context

CheXNet (Rajpurkar et al., 2017) — the seminal paper establishing DenseNet-121 for chest X-ray classification — reported 0.841 macro AUC-ROC on a comparable split of this dataset. CheXVision-ResNet, a custom residual SE-network trained from scratch with no pretrained weights, reaches 0.8141 macro AUC-ROC — within 0.027 of this benchmark. See the CheXVision demo for live inference, or the presentation deck for the project walkthrough.

Citation

@misc{chexvision2026,
  title={CheXVision: Dual-Task Chest X-ray Classification with Custom CNN and DenseNet-121},
  author={BIG D(ATA) Team},
  year={2026},
  howpublished={\url{https://huggingface.co/arudaev/chexvision-scratch}}
}

Downloads last month: -; Downloads are not tracked for this model. How to track

arudaev
/

chexvision-scratch