Oxipit at ECR: Save rads from burning out and ‘behind the scenes’ look at ChestEye

AI-based medical imaging startup Oxipit will showcase the ChestEye imaging suite at European Congress of Radiology in Vienna. The suite was recently awarded a CE mark which paves the way for clinical software deployment in 32 European markets. Oxipit team will also present two research papers in the Artificial Intelligence and Machine Learning sessions which outline the ‘behind the scenes’ development of it’s award winning suite.

The research will be presented by Oxipit’s Chief Medical Officer Naglis Ramanauskas, MD. Mr. Ramanauskas is Co-Founder of Oxipit medical imaging company.

The session titled “Deep learning based chest x-ray whole-image search and retrieval” details the benefits of back-catalogue chest X-Ray image search. A typical hospital possesses more than a decade worth of digital radiological images with their radiological descriptions in the PACS/HIS system. Deep Learning algorithm was used to index the images. The indexing process takes into account not only the pathology, but also the localization of the pathology, and the overall features of the radiograph.

A radiology resident was presented with a set of 77 challenging radiology images and tasked with evaluating and writing reports for each image without any assistance. Each of the reference images was subsequently indexed by the search system, and used to retrieve 10 radiographs from the hospital database of more than 200`000 images. After each report he was asked to review the the returned search matches and modify the report if he found it necessary. The report was modified after inspecting search matches for 56/77 cases. The impression was modified after inspecting search matches for 50/77 cases. The differential diagnosis was expanded for 28/77 cases. In conclusion, utilization of back-catalogue image search allowed to better evaluate the diagnosis, make the diagnosis more precise and reassure the medical specialist of the initial diagnosis.

“As a practicing radiologist myself, I see this experiment as one of the best showcases of AIaided diagnosis in radiology. Recent studies indicate radiology has the 7th highest speciality with the risk of burnout, with nearly half of specialists reporting it in 2018. The number of X-Ray, MRI and CAT scans increased five-to-tenfold in the last decade. Therefore we focused development of ChestEye to streamline the radiologist work process, reduce the time spent on mechanic diagnosis description, offer a knowledgeable second-opinion in order to improve radiologist professional well-being,” Ramanauskas said.

The presentation titled “Towards a deep learning model for exhaustive chest X-Ray pathology classification” focuses on training ChestEye algorithms to localize and describe 75 radiological findings. For the experiment a data-set of 301 255 CXR images were used. The data was divided into a training set (70%), a testing set (20%), and a validation set (10%). A single convolutional neural network (CNN) based on Inception was used to construct a model for CXR pathology classification. Oxipit model achieved AUC value of 93%.

ChestEye is the first AI-based full workflow medical imaging suite to be certified by a CE mark. ChestEye imaging suite encompasses a fully automatic computer aided diagnosis (CAD) platform, a search module for discovering similar-looking chest X-rays in a given database and a patient prioritization solution.

Oxipit will be showcasing ChestEye at AI-6 booth in the AI area at X1. You may schedule a meeting with the team at this link.


Oxipit is a computer vision software startup specialized in medical imaging. With a team of award-winning data scientists and medical doctors, the company aims to introduce innovative Artificial Intelligence/Deep Learning breakthroughs to everyday clinical practice. Oxipit are the authors of CE certified multi-award winning ChestEye radiology imaging suite.

Around the web

The Department of Defense is gifting Case Western researchers a grant to study the use of AI in determining whether patients require surgery. 

This study brings light to the prospect of a "fully automated solution" for echocardiogram analysis, experts reported.

Experts noted a "significant" reduction in false positives and false negatives using their modified machine learning model.