AI Assistance System
With RAIDAS we developed a computer-assisted system that uses artificial intelligence (AI) methods to automatically examine MRI scans for suspicious lesions and present them to the radiologist for further evaluation.
About A.I.gnostikum
At A.I.gnostikum, our goal is to reduce the workload for radiologists while simultaneously increasing diagnostic accuracy.
Our Technology
RAIDAS represents the forefront of AI-driven diagnostic tools. It utilizes neural networks trained on manually classified MRI time series to determine BI-RADS (for breast) and PI-RADS values (for prostate), providing radiologists with precise and reliable classifications as benign or malignant. The system’s key features include:
3D MRI Data Analysis
Tackling the complexities of volumetric MRI data through advanced preprocessing and customized network structures.
Explainable AI
Offering transparency in AI decision-making by highlighting critical areas in MRI data that influence classification results.
Seamless Integration
RAIDAS integrates smoothly into existing diagnostic environments.
Enhanced Diagnostic Confidence
Assisting radiologists by providing additional insights, thereby reducing the risk of oversight and improving diagnostic confidence.
Team
Dr. Manfred Prantl
Shareholder, CEO of AIgnostikum GmbH. He holds a doctoral degree in Computer Vision and Image Processing from the Technical University of Graz.
Prof. Dr. Dieter Szolar
Shareholder, Radiologist and consultant in medical matters; Head of the Diagnostikum radiology institute.
Specific Challenges in 3D MRI Data Analysis with AI
The development of an automated AI system for determining BI-RADS/PI-RADS values is based on neural networks that use manually classified MRI time series as input data.
The challenges are in analyzing 3D timeseries data as opposed to 2D image data, which resulted in larger data volumes and the need for adapted network structures. To address this, special preprocessing steps such as pre-segmentation and data augmentation were used, and customized network structures were developed.
Partners
Our development partners and sponsors support us in realizing our vision:
FFG
The development of RAIDAS has partly been supported by the Austrian Research Promotion Agency (FFG) under grant FO999908298.
Our Impact
RAIDAS has already demonstrated its effectiveness in clinical settings such as Diagnostikum Graz, where it processes thousands of MRI studies with an average prediction time of just 3 minutes per study. By consistently delivering accurate BI-RADS/PI-RADS scores, RAIDAS has become an invaluable tool for radiologists, contributing to more precise and timely diagnoses.
Schedule a software-demo appointment.
Contact us to schedule a free appointment!
