VinDr-CXR: An open dataset of chest X-rays with radiologist’s annotations

The dataset includes more than 100,000 chest X-ray images collected from major hospitals in Vietnam. From this raw data, the study shared 18,000 images annotated by 17 experienced radiologists. The dataset is divided into a training set of 15,000 images and a test set of 3,000 images. VinBigData designed and built a labeling platform for…

VinDr-RibCXR: A Benchmark Dataset for Automatic Segmentation and Labeling of Individual Ribs on Chest X-rays

Research and build a labeling dataset to develop algorithm for segmenting and noting rib arcs. Specifically, after collecting from 245 scans, chest X-ray images were partitioned and labeled with 20 different bone arches. This dataset, then, is the basis for training today’s most advanced deep learning algorithms (such as U-Net, FPN, U-Net++). Combining a standard…

DICOM Imaging Router: An Open Deep Learning Framework for Classification of Body Parts from DICOM X-ray Scans

The deep learning system, called DICOM Image Router, allows automatic classification of X-ray image types according to shooting indications. The system acts as an imaging router to precisely identify X-ray films by part or order, before they are sent to a specific diagnostic AI model.

Learning to Automatically Diagnose Multiple Diseases in Pediatric Chest Radiographs Using Deep Convolutional Neural Networks

The deep learning model is capable of diagnosing common pathologies on chest X-ray images in pediatric patients. Experiments show that the model achieves superior performance compared to existing methods. Especially, after this study, for the first time, a large-scale pediatric X-ray dataset (over 5,000 cases), was labeled and a new deep learning algorithm was introduced…

VinDr-SpineXR: A deep learning framework for spinal lesions detection and classification from radiographs

VinDr-SpineXR allows classifying between normal and abnormal spine X-ray images with 90% accuracy on the AUC measure (area under the curve indicates the ability to distinguish between patients with disease and not sick). In addition, the model is also able to accurately localize 06 types of common spinal injuries in Vietnam, including: bone spurs; disc…