Vojtěch Novotný, a student at the Faculty of Information Technology of the Czech Technical University in Prague (FIT CTU), and faculty alumnus Ing. Vojtěch Bešťák have won Rakathon 2026 with their team. This national hackathon focuses on the use of artificial intelligence in cancer prevention, diagnosis, and treatment. The team took first place among 65 projects with its prototype, KomPACS, designed to help healthcare professionals locate previous medical imaging examinations and obtain important information to support treatment decisions for cancer patients.
Before CT scans, magnetic resonance imaging (MRI), or PET scans, as well as before multidisciplinary team meetings, doctors often need to compare current findings with a patient's previous images. However, these images may not be available in the hospital's Picture Archiving and Communication System (PACS), particularly if the examination was performed at another healthcare facility. Locating them can then require numerous phone calls and delay preparations for further care.
Rakathon addressed this very problem through the PACS Comparator Check challenge, set by Motol University Hospital and Homolka Hospital. The winning team, 404: tumor not found, consisting of Vojtěch Novotný (FIT CTU), Vojtěch Bešťák (FIT CTU alumnus), Adam Karas (First Faculty of Medicine, Charles University), and Pavel Kos, responded with KomPACS (PACS Comparator). The project's aim is to automate the search for previous examinations and simplify the process of obtaining missing imaging data.
When a medical examination request is received, KomPACS searches available records in hospital systems and analyses the text of medical reports.
“Looking ahead, we also plan to search for relevant examinations in health insurance company data as soon as legislative changes make this possible. The current structure of the Czech healthcare system does not yet allow such data sharing, even though it would significantly improve coordinators' work and enhance the quality of patient care,” explains Vojtěch Bešťák.
Using artificial intelligence, the system assesses whether the images it finds may be relevant to the planned examination or to decisions made by a multidisciplinary team. If images are missing, it prepares the necessary information to request them. The final decision, however, remains with healthcare professionals. Artificial intelligence serves as a support tool, helping coordinators identify the required data more quickly and prepare the next steps.
The project's value was demonstrated through testing on 40 sample cases. In 60% of them, a relevant previous image was missing, even though it could have been important for assessing the patient's current condition. According to the team's findings, these images could not be located through conventional searches in hospital information and imaging systems because references to them appeared only in the free-text sections of medical reports.
“We want doctors to be able to focus on patients rather than chasing up images over the phone. KomPACS is designed to make this work significantly easier, and with future integration of insurance company data, it could also find examinations performed at other facilities,” says Vojtěch Bešťák.
“Our goal is to take KomPACS from a prototype to a solution that can operate in everyday hospital settings. That is why we are building on established HL7 v2 and FHIR standards and the InterSystems IRIS for Health platform. Before it can be tested in practice, however, we need to integrate it securely with the systems hospitals already use,” adds Vojtěch Novotný.
The project was developed during Rakathon 2026, held simultaneously in Prague, Brno, and Ostrava from 16 to 18 October. Nearly 500 participants, including students, developers, data analysts, doctors, and other professionals, spent 24 hours developing technological solutions to specific challenges in cancer care in the Czech Republic.
Learn more about the project, how it works, and its future plans