Digital AI platform for radiologists

Use case

Cognexa partnered with University Hospital in Central Europe on a project to establish a digital environment for utilizing artificial intelligence (AI) in radiology. The first impulse came from a need to was to develop an AI model for classifying X-ray images to detect the presence of Covid-19.

The vision was to have a platform using which the medical researchers and practitioners would be able to perform multidisciplinary research without the explicit dependance on programmers – namely being able to develop, study and evaluate their own AI models without the need for any coding skills.

Solution

Cognexa developed a comprehensive solution consisting of two main components. We used and customized the MONAILabel environment with combination in visualization software for radiologists – Slicer to enable advanced AI analysis of radiological images. The solution provides a platform for selecting most known segmentation models that can be trained on custom data.

To demonstrate the capabilities of the solution, Cognexa provided a trained open-source model capable of predicting Covid-19 presence in X-rays. Second, Cognexa built a web-based dashboard to facilitate user interaction with the custom data, to organize them into datasets ready for the annotation and training in the developed solution. The dashboard allows authorized users to browse the created datasets. The solution allows them to request predictions on selected images, view prediction history, and manage user accounts.

Result

The solution enabled the client to effectively utilize AI for analyzing radiological images. The custom AI model achieved a high level of accuracy in predicting Covid-19 presence in X-ray images. The user-friendly dashboard streamlined the process of requesting and viewing predictions, preparing datasets, utilizing AI in diagnostic processes, enhancing the overall workflow for the staff. This project showcases the successful application of AI in a healthcare setting, demonstrating its potential to improve diagnostics and patient care.