---
license: cc-by-4.0
size_categories:
- n>1K
language:
- en
tags:
- medical
- CT_SCAN
- CT-Scan
- Image
- Dicom
- Dataset
- Healthcare
- Sample
- Large
- Benchmark
- DeepLearning
---

**Dataset Description:**

**This dataset is a large-scale collection of CT (Computed Tomography) scan reports without clinical findings, containing data from 276,113 patients and 65,303,386 medical images, designed to support the development and training of advanced healthcare AI, medical imaging, and diagnostic AI systems.**

The dataset captures authentic imaging characteristics such as scanner variability, acquisition protocols, and patient positioning, along with structured and unstructured clinical narratives. This makes it highly valuable for building accurate, scalable, and production-ready AI systems.
Additionally, this dataset can be used in pipelines for **Supervised Fine-Tuning (SFT) workflows**, especially for enhancing model sensitivity and calibration.

![Screenshot 2026-04-13 152720](https://cdn-uploads.huggingface.co/production/uploads/693ab313ff1770594f99afee/6fpMz5q-jq14aMFikl9Un.png)

**Dataset Specification**

    -Patients: 276,113
    -Images: 65,303,386
    -Modality: CT (Computed Tomography)
    -Type: Medical images without clinical findings
    -Data Source: Clinical CT scan reports
    -Body Regions: Brain, Chest, Abdomen, Spine, etc.
    -Data Nature: Real-world clinical data

**Key Use Cases**

    -Baseline learning for diagnostic AI models
    -False positive reduction
    -Medical image segmentation
    -Radiology report generation
    -Model validation and calibration

**Value of This Dataset**

    -Enables learning of normal anatomical structures
    -Improves model specificity and reduces false alarms
    -Supports binary and multi-class classification tasks
    -Essential for anomaly detection systems
    -Helps build clinically safe and reliable AI solutions
    -Supports real-world healthcare deployment

**Basic JSON Schema**

```json
{
  "patient_id": "string",
  "image": "image",
  "series_number": "string",
  "study_uid": "string",
  "series_uid": "string",
  "instance_number": "int32"
}