Title: Analysis of Equinal Szintigraphy Images

Project duration: 2022 – 2026

Research Area: Life Science and Medicine, Veterinary Medicine

The manual evaluation of scintigraphy images is time-consuming, subjective, and requires substantial expertise. Machine learning (ML) offers the potential to streamline and accelerate this process, while also increasing objectivity and consistency.

This project focuses on two clinically relevant anatomical regions:

• Cervical Spine C6/C7 (HWS C6/C7):
  Pathologies in the caudal cervical spine, particularly around the C6/C7 facet joints, often show increased radiotracer uptake. This can make it difficult to interpret pathological changes in scintigraphy images.
  The gold standard for diagnosis is the measurement of radiation intensity differences using regions of interest (ROI). However, this method is especially complex in this area and requires considerable experience.
• FTU (Fesselträgerursprung / Origin of the Suspensory Ligament):
  Suspensory ligament desmitis is a common pathology in sport horses.

Aims

• Develop an AI-based system for automated evaluation of scintigraphy images
• Improve objectivity and reproducibility in diagnostic results
• Detect inflammation in specific anatomical regions
• Significantly reduce image evaluation time
• Improve diagnostic performance and enable clinical deployment

Problems

• Manual image evaluation takes over 1 hour per image
• ROI-based analysis shows relatively low accuracy (~60%)
• Heterogeneous data sourced from three clinics (Bargteheide, Isernhagen, Leipzig)
• Need for comparison between traditional ROI analysis and AI-based model results

Technology

• Data Preprocessing
• Image Selection using Convolutional Neural Networks (CNN)
• Data Augmentation to increase model robustness
• Image Masking with U-Net for anatomical segmentation
• Labeling using Random Forest classifiers
• Sequence Smoothing with custom algorithms
• Final Classification using CNN-based models

Outlook

• A Living Lab Demonstrator in the form of a web application is planned

Team

Lead

• Dr. Christian Martin (Database Group Leipzig – ScaDS.AI Dresden/Leipzig)

Team Members

• Anja Neumann (Applied AI and Big Data – ScaDS.AI Dresden/Leipzig)
• Sigita Krüger (Clinic for Horses – Veterinary Medicine Leipzig University)
• Prof. Kerstin Gerlach (Imaging Diagnostics – Veterinary Medicine Leipzig University)

Funding

• Federal Ministry of Research, Technology and Space; Free State of Saxony