DPU - Symposium 2022

Medica l Image Ana lysis & Ar t i f icia l Intel l igence Symposium 2022 Nr. 15: Ass.-Prof. Dr. Sepideh Hatamikia &Dr. Florian Schwarzhans, PhD, MSc: Evaluation of radiomics robustness on mammograms using different modifications of manually drawn regions of interest Curriculum Vitae: Ass.-Prof. Dr. Sepideh Hatamikia completed her PhD and also a postdoctoral position in biomedical engineering field at Medical University of Vienna at the Department of Medical Physics and Biomedical Engineering at the Digital Image Processing Laboratory. Since march 2022, she joined Danube Private University (DPU) and currently she is the Head of Computational Imaging in the Medical Image Analysis and Artificial Intelligence (MIAAI) research group. In addition, Ass.-Prof. Dr. Hatamikia is a researcher at Austrian Center for Medical Innovation and Technology (ACMIT) since January 2018 and continues collaboration with this research center on different research projects. Curriculum Vitae: Florian Schwarzhans holds an MSc degree in Medical Informatics from the Medical University of Vienna. He has a background in informatics with a focus on programming using C, C++, MatLab and Python, as well as a background in electronics specializing in biomedical engineering. His research interests include medical image processing with a special focus on automatic graph-based segmentation algorithms, deep learning methods for both image classification and segmentation, and the development and implementation of parallel algorithms for medical image processing and analysis using CUDA. He has developed real-time retinal tracking software for a prototype PS-OCT system and software for fast parallel reconstruction of OCT volumes from raw data via the GPU, which is actively being used in multiple OCT systems. Abstract: Evaluation of radiomics robustness on mammograms using different modifications of manually drawn regions of interest Abstract: Radiomics-based models have shown promise in detecting and characterizing breast lesions and in predicting outcome. However, radiomics features show limited robustness to variations in lesion delineation but their selection according to their robustness may facilitate the clinical translation of radiomics-based predictors. In this study, we investigate how different mathematical region of interest (ROI) modifications and tumor size affect radiomics features. In addition, what is the relationship between tumor ellipticity and feature robustness. This is the first systematic analysis on variations of ROIs outlining breast cancer on mammograms.