Platform for Automated Analysis of Medical Images Based on Artificial Intelligence

  • Score Awaiting client review
  • Date Published
  • Reading Time 4-Minute Read

How can you use machine learning for analyzing medical images? Fayrix’s medical platform development case study shows the ways to do it effectively.

Client Background:

Binomix LLC was created on a basis of the National Research Lobachevsky State University. For now, Binomix is engaged in the production of a digital X-ray detector and the development of artificial intelligence programs for processing the results of medical imaging studies.

Since 2019, the company has become the official distributor of RADLogics and uses its Virtual Resident product at the heart of BinomixAI. The main technology is the decoding of medical images for the purpose of diagnosing and detecting pathologies. It allows the processing of an incoming image and provides the doctor with a report, which contains a snapshot of the pathologies found, their size, volume and structure.

Product Overview

Binomix.connect is an integration platform that enables automated analysis of medical images based on artificial intelligence. The platform implements continuous improvement of diagnostic parameters through machine learning.

The system consists of 7 modules:

Distribution AI controller: The controller verifies the study and validates it against a specific set of metadata.

Diagnostic Material Controller: This module implements parsing of the metadata of incoming studies, brings the data into a single format, and provides data sending to the AI module to obtain results.

Report Configurator: This module provides a web interface that allows you to edit the content of PDF reports. The configurator allows you to create reports in various formats and independently select the required values.

Centralized billing: A module that allows you to track in detail quantitative indicators at all levels of the system.

DICOM database: Object storage that works with DICOM files. Includes a status system for files located in it.

Report generator: This module generates a PDF report with a contoured image containing the results of processing the AI module.

AI modules: One or more AI modules that process research and generate preliminary results.

Business Challenge

Binomix came up with the idea of creating a revolutionary medical platform of analyzing tens of thousands of X-rays per day, giving instant results. For example, in cases of Covid-19, it shows the area of lung damage, percentage of damage, and possible risks. At the same time, Binomix had difficulties in building a high-class team, meeting tight deadlines and uncertainty with the roadmap and the whole product concept realization.

  1. Constantly changing product requirements with tight deadlines;
  2. The lack of prior product research;
  3. The need for engineers with a specific background and special characteristics, ready to solve unique architectural problems.


Fayrix formed an initial team for the Binomix project in just a couple of weeks with a high-quality Project Manager on the Fayrix side.

Team Building Steps: 

  • Search for high qualified Project Manager on the Fayrix side.
  • Formation of the initial team for the project.

Task Trackers vs Communication Tools

The team has become a fully-fledged Binomix unit working by the FTE model, which was managed by the project manager on the Fayrix side. The work process is based on the usual for the client communication environment.


  • Bitbucket
  • Telegram
  • Google sheets

Software Infrastructure:

  • Jira – task tracker for internal tasks

Hardware Infrastructure:

  • Amazon servers

Project Results: Team Deliverables

A team of 6 qualified developers has accelerated the development of the main product by 15%. For now, the platform allows:

  • to set up a seamless interaction of medical IT systems with AI algorithms;
  • to improve existing AI algorithms for the analysis of medical images;
  • to create a report template for a doctor describing an image;
  • to work with digital X-ray images obtained on various radiation diagnostic devices (CT, PET-CT and X-ray images).

Team Responsibility Area

Back-end development:

  • Server-side based on PHP
  • Debugging work
  • Unit testing


  • Load testing
  • Functional testing
  • Calibration testing

Sprint Tasks

At the moment sprint contains an average of 60 tasks which allows the team to follow Binomix’s backlog and issue releases every 2 weeks.

During our partnership, the Fayrix team has integrated the Binomix platform with the Moscow services of the Health Department. At the moment, the platform can simultaneously analyze hundreds of thousands of X-rays and generate an instant report on who is sick and to what extent, and who is definitely not.

How We Did It:

  • 2-week sprint.
  • Planning all upcoming work for the team every month.
  • Constant cooperation with the client.

Sprint Results

Binomix has released an integration platform that enables automated analysis of medical images based on artificial intelligence. Fayrix’s high-quality team completed the assigned tasks in a short time and increased the productivity and quality of the developed product by up to 20%.