PhD Scientific Days 2024

Health Sciences III.

The role of X-ray-based radiomics in diagnosing post-COVID patients

Előadó neve

Dr. Fülöp, Andrea

Neptun code

IUDZQC

Előadó munkahelye

National Korányi Institute for Pulmonology

Előadó telefonszáma

+36300124050

Előadó e-mail címe

andreafulop94@gmail.com

Az előadás címe

The role of X-ray-based radiomics in diagnosing post-COVID patients

Szerző(k) neve és munkahelye

Andrea Fülöp¹, Jeovanis Gil², Anita Horváth-Rózsás³, Diana Solymosi¹, Ferenc Rényi-Vámos⁴, Krisztina Bogos¹, Bence Ferencz⁵, Judit Berta¹, György Marko-Varga², Balázs Döme⁶, Anna Kerpel-Fronius¹, Zsolt Megyesfalvi⁷

1: National Korányi Institute of Pulmonology, Budapest, Hungary
2: Clinical Protein Science & Imaging, Department of Biomedical Engineering, Lund University, Lund, Sweden
3: National Koranyi Institute of Pulmonology, Budapest, Hungary
4: National Korányi Institute of Pulmonology, Department of Thoracic Surgery, Semmelweis University and National Institute of Oncology, National Institute of Oncology and National Tumor Biology Laboratory, Budapest, Hungary
5: National Korányi Institute of Pulmonology, Department of Thoracic Surgery, Semmelweis University and National Institute of Oncology, Budapest, Hungary
6: National Korányi Institute of Pulmonology, Budapest, Hungary; Department of Thoracic Surgery, Semmelweis University and National Institute of Oncology; Department of Thoracic Surgery, Comprehensive Cancer Center Vienna, Medical University of Vienna, Vienna, Austria; Department of Translational Medicine, Lund University, Lund, Sweden
7: National Korányi Institute of Pulmonology, Budapest, Hungary, Department of Thoracic Surgery, Semmelweis University and National Institute of Oncology, Department of Thoracic Surgery, Comprehensive Cancer Center Vienna, Medical University of Vienna, Vienna, Austria

Bemutatás módja

Szóbeli

Szekció

Health Sciences III.

Language of the presentation

English

Preferred session

Health Sciences

Összefoglaló szövege

Introduction: CT- and chest X-ray (CXR)-based artificial intelligence algorithms effectively distinguish COVID-19 pneumonia from other types of pneumonia and aid in estimating the prognosis of the disease, there are currently no clear radiomic features for predicting long-term symptoms.

Aims: Our study aims to correlate radiomic features from X-ray images with clinical symptoms in long COVID patients.

Methods: In our study, we examined the patient population of the OKPI, consisting of 260 acute COVID-19, 278 post-COVID, and 435 non-COVID patients. The first step in the bioinformatic analysis of radiological images was a four-layer segmentation based on artificial intelligence. Subsequently, the cardiothoracic ratio (CTR) was determined for each image. The determination of important radiomic features and the development of neural networks suitable for the classification of different patient groups were based on machine learning algorithms.

Results: The algorithm accurately identified the lungs, heart, and background. Elevated CTR was observed in 4.84%, 6.27%, and 23.63% of cases in non-COVID, post-COVID, and COVID-19 patients, respectively. Elevated CTR was more frequently observed in post-COVID patients compared to non-COVID individuals, no significant difference (p=0.448) was found in the distribution of CTR among the mentioned patient groups. The developed diagnostic algorithm was able to differentiate the three study groups with an accuracy between 65-69%. The differentiation of groups was based on 96 radiomic parameters, of which 28 played a decisive role in decision-making. Visualization models using Salinecy map and GradCAM++ techniques were created to verify the trained convolutional network.

Conclusions: Analyzing post-COVID CXR in everyday practice is often challenging for radiologists. Artificial intelligence-based algorithms may offer promising solutions for recognizing pathological patterns invisible to the human eye.

Funding:
- BD, ZM: Hungarian National Research, Development, and Innovation Office (2020‐1.1.6‐JÖVŐ, TKP2021‐EGA‐33, FK‐143751 and FK-147045)
- ZM: New National Excellence Program of the Ministry for Innovation and Technology of Hungary (UNKP‐20‐3, UNKP‐21‐3 and UNKP-23-5), Bolyai Research Scholarship of the Hungarian Academy of Sciences, International Lung Cancer Foundation Young Investigator Grant 2022

University

Semmelweis University

Supervisor

Dr. Megyesfalvi Zsolt PhD

Publication of my abstract

I do not give consent to the publication of my abstract on the website of the congress.

Kind

Szabad

Status

elfogadva

Accepted presentation method

szóbeli

Előadás fájl jóváhagyás

nem rendelkezett róla

Előadó

8115

Start

14:45

End

14:55