publications

2025

1. Myocardial perfusion imaging SPECT left ventricle segmentation with graphs

   Ádám István Szűcs, Béla Kári, and Oszkár Pártos

   EJNMMI Physics, Mar 2025

   Abs DOI

   Various specialized and general collimators are used for myocardial perfusion imaging (MPI) with single-photon emission computed tomography (SPECT) to assess different types of coronary artery disease (CAD). Alongside the wide variability in imaging characteristics, the apriori “learnt” information of left ventricular (LV) shape can affect the final diagnosis of the imaging protocol. This study evaluates the effect of prior information incorporation into the segmentation process, compared to deep learning (DL) approaches, as well as the differences of 4 collimation techniques on 5 different datasets.

2024

1. Continuous Max-Flow Augmentation of Self-Supervised Few-Shot Learning on Spect Left Ventricles

   Ádám István Szűcs, Béla Kári, and Oszkár Pártos

   In 2024 IEEE International Symposium on Biomedical Imaging (ISBI), Mar 2024

   DOI
2. Self-supervised segmentation of myocardial perfusion imaging SPECT left ventricles

   Ádám István Szűcs, Kristóf Horváth, Katalin Sólyomvári, and 2 more authors

   In Proceedings of the 2023 10th International Conference on Bioinformatics Research and Applications, , Barcelona, Spain, , Mar 2024

   Abs DOI

   Myocardial Perfusion Imaging (MPI) is the golden standard in detecting and assessing Coronary Artery Disease (CAD). Due to its role and incidence in diagnostics, automated and semi-automated solutions are available aiding diagnoses based on the perfusion and function of the left ventricle. Evaluation with Single-Photon Emission Computed Tomography (SPECT) cameras often takes > 30 minutes and due to patient cardinality, shortened acquisition produces images with lower Signal-to-Noise Ratio (SNR), challenging automated solutions. Here a fully automated method is presented, solving the left ventricle (LV) segmentation problem. The method is a 3D U-Net-based, self-supervised learning (SSL) approach, with a rich augmentation pipeline. A brief study is presented where the proposed algorithm is compared to the current state-of-the-art solution in Machine Learning (ML), with the evaluation of various segmentation metrics of the left ventricular muscle tissue. The proposed method showed an effective 89 % precision on the segmentation task of MPI studies on both simulated and real datasets. In conclusion, the technique can aid the evaluation of MPI studies in clinical environments by automating the segmentation of studies with a high success rate.

3. Projection geometry invariant nonrigid motion correction

   Ádám István Szűcs, Béla Kári, and Oszkár Pártos

   In 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Mar 2024

   DOI

2019

1. Automated non-linear motion detection and estimation for MPI SPECT

   Ádám István Szűcs, and Béla Kári

   In 2019 1st IEEE International Conference on Gridding and Polytope Based Modeling and Control (GPMC), Mar 2019

   DOI
2. Improving graphics programming with shader tests

   Ádám István Szűcs

   Pollack Periodica, Mar 2019

   DOI

2016

1. Active contours for selective object segmentation

   Jozsef Moinar, Adam Istvan Szucs, Csaba Molnar, and 1 more author

   In 2016 IEEE Winter Conference on Applications of Computer Vision (WACV), Mar 2016

   DOI

2014

1. Distributed computing of simultaneous Diophantine approximation problems

   Norbert Tihanyi, Attila Kovács, and Adám Szucs

   Stud. Univ. Babes-Bolyai Math, Mar 2014