AI detects pathologies in thorax X-rays and validates results with medical reports
A research team from the Universitat Jaume I of Castelló and the Foundation for the Promotion of Health and Biomedical Research of the Valencian Community (Fisabio), with the collaboration of the Radiodiagnostics Service of the Castelló General University Hospital(HGUCS), has developed a tool for the automatic detection of pathologies in thorax X-rays based on the experience and expertise of radiology staff that are validated from the automatic analysis of medical reports.
The new intelligent system to support radiodiagnosis, based on artificial intelligence (AI) computing techniques and the processing of large volumes of data (big data), aims to identify urgent pathologies in chest X-rays in adults and thus prioritize the preparation of the report by the radiology specialist, which would allow the implementation of a radiological triage system. This is expected to achieve a more agile management of urgent cases coming from the Emergency Department of the HGUCS.
"Thissoftware combines artificial intelligence and big data to identify common pathologies in simple X-rays", says Manuel F. Dolz Zaragozá, researcher of the Gen-T Programme of Excellence of the Generalitat Valenciana and member of the High-Performance Computing and Architectures research group (HPC&A) of the Universitat Jaume I.
In the opinion of Katty Delgado Barriga, radiologist at the Castelló General University Hospital and researcher at the Fisabio Foundation, "in this way, medical staff can prioritize the care of patients with more serious pathologies and reduce the waiting time for patients with simpler radiological studies".
This software consists of a module for the detection of pathologies in thorax X-rays, which can be extended in the future with other modules for the identification of pathologies in other types of medical images. To do this, the system receives anonymous images from medical centers and responds with a prediction of the pathologies found. The collaboration of radiology professionals in its development has allowed them to use their knowledge and experience to reinforce the learning of the artificial intelligence algorithms used.
This new tool aims to prioritize the interpretation of studies with critical or urgent findings, which will optimize the triage of X-rays to reduce the waiting time for each report and thus streamline the intra-hospital work circuit at the Castelló General University Hospital.
Currently, the research team has already completed the proof of concept and has started the validation phase in a real hospital environment. The registered software is aimed at both teleradiography process companies and hospital radiology departments, and is currently in the process of seeking a licensing agreement for exploitation.
Provided by Asociacion RUVID