Case Study: Aortic CT 3D Segmentation for Radiology AI

Project Overview

Nurea, a medical AI company specializing in radiology, collaborated with EGY-I Data Annotation on a 3D aortic CT segmentation project to generate high-quality annotated datasets for radiology AI.

The objective was to generate annotated datasets to support the training of AI models focused on vascular analysis.

This project required careful handling of anatomical structures and consistency across volumetric data.

Dataset Scope

Project Statistics

Total CT scans segmented: 50
Average slices per scan: ~500
Total slices annotated: ~25,000
Validation views: Axial, Coronal, Sagittal

Each scan was analyzed and validated across all three planes to ensure complete volumetric consistency and correct anatomical continuity.

Segmentation Targets – Primary Anatomical Structures

The following anatomical structures were segmented:

Ascending aorta
Aortic arch
Descending thoracic aorta
Abdominal aorta

Annotated Structures

The following structures were annotated:

Aortic lumen
Thrombus
Calcifications
Other vascular features depending on scan characteristics

Annotation Workflow

1. Data preparation

CT preprocessing, orientation validation, slice alignment verification.

2. AI-assisted initialization

AI-assisted initialization to accelerate segmentation.

3. Expert manual segmentation

Expert manual segmentation by trained medical annotators using tools such as 3D Slicer and ITK-SNAP.

4. Multiplanar validation

Multiplanar validation across axial, coronal, and sagittal views.

5. Final model consistency check

Final model consistency check through 3D inspection.

Our Expertise in High-Precision Medical Segmentation

EGY-I Data Annotation specializes in complex medical imaging annotation and works with AI companies developing radiology models worldwide.

Our strengths include:

✔ Annotation performed by medical professionals and trained medical annotators

✔ Extensive experience with CT, MRI, CBCT, ultrasound, and angiography datasets

✔ Expertise in fine anatomical structures such as vessels, nerves, and small lesions

✔ Ability to handle large-scale datasets with tens of thousands of slices

✔ Strong experience in vascular segmentation and pathology annotation

Quality Control Process

We implement a multi-layer QC system to ensure clinical-grade dataset quality.

Level 1 – Annotation Review

Initial review by senior medical annotators to verify segmentation accuracy and boundaries.

Level 2 – Clinical Validation

Focused review to ensure anatomical and clinical correctness.

Level 3 – Dataset Consistency Audit

Final review to verify cross-scan consistency, class labeling accuracy, and segmentation continuity across slices.

This structured QC pipeline significantly improves dataset reliability and directly contributes to better AI model performance.

Project Results

Delivered Outcomes

✅ 50 CT scans segmented
✅ ~25,000 slices annotated
✅ High anatomical accuracy across 3D volumes
✅ Dataset delivered for AI usage

Images

Impact

Accurate segmentation of vascular structures such as the aorta is critical to train AI systems intended for cardiovascular domains.

By combining medical expertise, advanced annotation tools, and rigorous quality control processes, EGY-I Data Annotation helps medical AI companies build high-quality datasets that accelerate the development of reliable clinical AI solutions.