ALISON Slovakia
AI-translated
Where

cybersecurity meets artificial intelligence

ALISON Slovakia brings to the partnership a solid corporate foundation and extensive technical and legislative cybersecurity experience proven in practice.

In October 2026, ALISON Slovakia becomes a minority co-owner of Gauss Algorithmic.

Our cybersecurity division has been working with the Gauss team for two years, and the agreement was preceded by a careful selection of partners on both sides.

Gauss Algorithmic is based in Brno and is one of the leading Czech companies in artificial intelligence development, data analytics and the integration of machine learning with DevOps development. It has been on the market since 2013 and has won several international awards.

The references of Gauss Algorithmic include a long list of clients from logistics, banking, healthcare and telecommunications. The success of the company is built on a team of professionals, its own research, infrastructure and platform.


Together we will overcome one of the current obstacles to research – how to use valuable data while protecting sensitive information, patient identity and research know-how. From data preparation and pseudonymisation through secure sharing to the training of AI models.

Entering the AI era with our own solutions

The joint team of ALISON Slovakia and Gauss Algorithmic is developing an original solution that benefits both medicine and the training of AI models, with an emphasis on protecting sensitive data at every step.


DICOM (Digital Imaging and Communications in Medicine) is a standard for storing and exchanging medical images, for example from CT, magnetic resonance or ultrasound. Besides the image, the file also contains accompanying data about the patient and the examination.

Sanitex

A software solution for local anonymisation and pseudonymisation of medical imaging data in the DICOM format.

Goal

To increase patient privacy protection and enable safer use of medical data for research, education, analytics and development while preserving its clinical and technical value.

What matters is whether, before use in third-party AI or outside the healthcare facility, identifying data has been reliably removed from DICOM, including from the images, and whether the result of the anonymisation has been checked.

Sanitex makes it possible to identify and remove or replace personal and sensitive data contained in the metadata as well as in the image content itself.

Output

A comprehensive AI-driven solution deployed locally in the infrastructure of the healthcare provider, with no need for an internet connection. Sanitex covers both key aspects of DICOM anonymisation: de-identification of metadata as well as detection and masking of sensitive information written directly into the pixel data of images, so-called burned-in PHI.


Key features of Sanitex

Sanitex uses a two-layer approach.

Rule-based layer according to PS3.15

  • Anonymisation of more than 40 known PII tags in line with the standards
  • Processes patient names, IDs, dates, UIDs, physician names, institution details and all vendor-specific private tags
  • This layer is deterministic and complete

AI-powered layer – LLM model

  • A local LLM reads the actual content of DICOM fields to catch PII that the rules miss, for example patient names entered in clinical descriptions
  • The AI is DICOM-aware and multilingual

  • Modular pipeline architecture
  • Use of pre-trained models LLM/NER for personal data classification
  • Elimination of the need to train on sensitive medical data
  • Tested and optimised for NVIDIA DGX Spark