PP.7.2 INDUCTION OF COMPLEX DNA DAMAGE AFTER PROTON THERAPY BEAM IRRADIATION TO PLASMID DNA AND HUMAN PROSTATE CANCER CELLS

S36 Abstracts / Physica Medica 127S1 (2024) S1–S82 identify informative combinations of features and can be combined with other clinical data. Neural Networks are applied to head & neck PET oncological data to build predictive diagnostic, and prognostic models in relation to clinical endpoints. iii) The utilization of such AI decision-aid models in medical imaging and oncology is hindered by their black-box nature. For such purpose an interactive web application is presented with an interpretable AI classifier on tumor malignancy to visualize the model’s explainability. iv) Regression models are developed for a predictive personalized dosimetry tool for pediatric patients on a web application for clinical use. v) The generation of AI-based PET data is investigated synthesizing heterogeneous tumors for head & neck PET data, that are further combined in MC simulations using digital human phantoms. Conclusion: The present work showcases AI-based solutions to optimize patient management, clinical treatment strategies and personalized predictive diagnostic and therapeutic protocols by developing novel methods to facilitate decision-aid systems. Physica Medica 127S1 (2024) 104610 https://doi.org/10.1016/j.ejmp.2024.104610 PP.6.7 MACHINE-LEARNING-BASED PRIORITISATION OF LONG COVID PATIENTS FOR SPECIALIST CONSULTATION A. Kerezis1,2, A. Antonoglou1, A. Asimakos1, N. Athanasiou1, S. Spetsioti1, K. Dalakleidi2, P. Asvestas3, P. Katsaounou1, S. Golemati1 1First Department of Intensive Care, Medical School, National and Kapodistrian University of Athens, Athens, Greece, 2School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece, 3Medical Image and Signal Processing Laboratory, Department of Biomedical Engineering, University of West Attica, Athens, Greece Background: Long COVID, a post-acute sequela of SARS-CoV-2 infection, poses a significant burden on healthcare systems. Efficient triaging of patients for specialist consultation is vital for timely and appropriate care. A machine learning (ML) algorithm is proposed to prioritise long-COVID patients for initial consultations with cardiologists, pulmonologists, or psychiatrists. Methods and Materials: A ML model was developed using a Support Vector Machine (SVM) classifier with Synthetic Minority Oversampling Technique (SMOTE) for class imbalance handling and feature selection techniques. Data from 175 patients included demographics, COVID-19 severity, vaccination status, validated questionnaires (BECK Depression, PTSD, ABC, CAT, EQ-5D, HADS Depression/Anxiety, SF-36), and specific symptoms (intestinal disorders, sleep disorders, precardial pain). The model was trained and validated using 5 fold Stratified K-Fold Cross-Validation and shuffling of the data. Results: The SVM model achieved an accuracy of 0.67 and an Area Under the Curve (AUC) of 0.84, outperforming other models (Random Forest, K-Nearest Neighbours, Decision Tree, XGBoost). Notably, the SF-36 General Health (SF-36 GH-N) score emerged as the most important factor for patient risk group classification. This suggests a patient’s overall health perception plays a crucial role in identifying the most needed specialist for their initial consultation. The algorithm categorises patients into risk groups based on predicted needs for cardiology, pulmonology, or psychiatry consultations. Conclusion: This study demonstrates the potential of an MLbased approach to prioritise long-COVID patients for specialist consultations. By leveraging patient data and validated questionnaires, the algorithm can guide resource allocation and expedite access to appropriate specialists, improving patient care. Future work will investigate the impact of the algorithm on clinical workflow and patient outcomes in a prospective study. Physica Medica 127S1 (2024) 104611 https://doi.org/10.1016/j.ejmp.2024.104611 Biophysics and Radiobiology (BRB) PP.7.1 MULTIFREQUENCY TYMPANOMETRY: FROM MIDDLE EAR MECHANICS TO DECISION MAKING IN EAR SURGERY C. Tsilivigkos1, A. Warnecke2, E. Ferekidis1 11st Department of Otolaryngology, Hippokration General Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece, 2Department of Otolaryngology, Medizinische Hochschule Hannover, Hannover, Germany Background: Multi-frequency tympanometry provides data on how the components of admittance (conductanceG, mass reactance, and stiffness reactance), vary in different frequencies. It also measures the resonant frequency of the middle ear, which is an indicator of alterations in sound transmission through the system. Materials and Methods: A comprehensive review of the current literature regarding the application of multifrequency tympanometry in the differential diagnosis of middle and inner ear diseases will be conducted. Results: Data regarding the utilization of multifrequency tympanometry as a diagnostic tool and evidence supporting the development of two otosurgical techniques in the 1st university clinic of Athens, namely chondrotympanoplasty and microtraumatic stapedotomy, will be provided. Conclusion: The current study will provide insight on the function of multifrequency tympanometry as a diagnostic tool based on middle ear mechanics. It will also highlight the significance of the technique in middle ear surgery planning and evolution. Physica Medica 127S1 (2024) 104612 https://doi.org/10.1016/j.ejmp.2024.104612 PP.7.2 INDUCTION OF COMPLEX DNA DAMAGE AFTER PROTON THERAPY BEAM IRRADIATION TO PLASMID DNA AND HUMAN PROSTATE CANCER CELLS M. P. Souli1,2, Z. Nikitaki1,2,3,4,5, E. Spyratou6, E. P. Efstathopoulos6, L. Sihver2,7, A. G. Georgakilas1 1DNA Damage Laboratory, Physics Department, School of Applied Mathematical and Physical Sciences, National Technical University of Athens, Athens, Greece, 2Atominstitut, Technische Universität Wien, Wien, Austria, 3ABTE UR4651, University of Caen Normandy, Caen, France, 4UMR6252 CIMAP, Team ARIA, Application in Radiobiology with Accelerated Ions, GANIL, Caen, France, 5Bâtiment Recherche, Centre François Baclesse, Caen, France, 62nd Department of Radiology, Medical School, National and Kapodistrian University of Athens, Athens, Greece, 7Nuclear Physics Institute of the Czech Academy of Sciences, Prague, Czech Republic Background: DNA damage studies are key to cancer radiotherapy, as they facilitate treatment optimization through comprehension of irradiation’s impact on cancer cells at molecular level, leading to more precise and effective tumor targeting while minimizing damage to surrounding healthy tissues. The typically studied Abstracts / Physica Medica 127S1 (2024) S1–S82 S37 biological systems are plasmids, mammalian DNA, or cells, while radiation qualities may include X-rays, protonor carbon ionbeams. Materials and Methods: This study aims to measure DNA damage in human prostate cancer cells (DU145) irradiated in the Spread-Out Bragg Peak (SOBP) region and plasmid pBR322 irradiated across a typical proton treatment plan at the MedAustron ion therapy facility (energy range 137-198 MeV and Linear Energy Transfer (LET) 1-9 keV/Mm). Our results highlight the crucial role of LET and Reactive Oxygen Species in DNA fragmentation, which was assessed using Agarose Gel Electrophoresis. Atomic Force Microscopy was utilized to measure apparent DNA lengths, providing insights into the impact of increasing LET on the induction of highly complex DNA damage. Results: Our combined methodologies revealed an increased level of DNA damage complexity at the SOBP and distal SOBP fall off. By dividing the mean rate of DSB induction for protons by the DSB induction rate for X-rays at the entrance of the beam, an apparent RBE value of 0.61 at the entrance, 0.49 in the SOBP plateau, and 0.18 at the SOBP fall-off was estimated. For human prostate cancer cells DU145, we detected a significant induction of complex damage at the distal SOBP fall-off, where highest LET is expected. Conclusion: Biological effects along a tumor area, based on the apparent RBE variable inferred from the present study, varies notably in contrast with the clinically utilized constant RBE of 1.1. This study highlights the limitation of constant RBE adoption and emphasizes the necessity for novel biomarkers to overcome existing limitations in complex DNA damage detection. Physica Medica 127S1 (2024) 104613 https://doi.org/10.1016/j.ejmp.2024.104613 PP.7.3 RAMAN SPECTROSCOPY FOR THE EVALUATION OF CANDIDA AURIS METABOLOMICS C. Petrokilidou1, E. Pavlou1, A. Velegraki2,3, A. Simou3, I. Marcellou3, G. Gaitanis4, N. Kourkoumelis1 1Department of Medical Physics, University of Ioannina, Ioannina, Greece, 2Medical School, University of Athens, Athens, Greece, 3Mycology Laboratory, BIOIATRIKI S.A, Athens, Greece, 4Department of Skin and Venereal Diseases, University of Ioannina, Ioannina, Greece Background: Candida auris is an emerging fungal pathogen that has been designated as an urgent and critical public health threat by both the United States Centers for Disease Control and Prevention (CDC) and the World Health Organization (WHO). This pathogen is associated with hospital outbreaks and exhibits a higher level of resistance to antifungal drugs compared to other Candida species. C. auris has been found to colonize not only the skin surface and hair follicles but also specific anatomical sites, including the nostrils, armpits, and inguinal folds. Materials and Methods: This study employed Raman spectroscopy to analyze the spectroscopic profile of three major pathogens, C. auris, C. parapsilosis, C. albicans. The strains were cultured on Dixon’s agar for 14 days at 32°C. Raman spectra were then directly obtained from the culture medium using a 785 nm laser (Coherent, USA), with two optical fibers (Thorlabs) facilitating the collection of scattered radiation. An f/1.3 Raman spectrometer (Wasatch Photonics) was used for spectral acquisition. The spectra were subsequently analyzed with a focus on the Candida-specific spectral region, and principal component analysis (PCA) was utilized for data interpretation. Results: PCA revealed discrimination among Candida species, primarily attributed to differences in the 1171 cm–1 and 1451 cm–1 bands in the Raman spectra. Variations in these peaks reflect differences in the metabolism of Dixon’s agar by the pathogens. Intensity ratios (I1450/I1170) for each Candida cluster, revealed that some strains exhibited low metabolic activity, whereas others displayed high metabolic activity. Conclusion: PCA demonstrated distinct differences between the Raman spectra of Candida species. Our study suggests that the ratio of the peaks presented in the Raman spectra of the pathogens could be a useful biomarker identifying the metabolic profile of Candida species, potentially aiding rapid diagnosis and treatment strategies. Physica Medica 127S1 (2024) 104614 https://doi.org/10.1016/j.ejmp.2024.104614 Biomedical engineering (BME) PP.8.1 RELAXOMETRY AND RELAXIVITY ASSESSMENT OF IRON OXIDEBASED MRI CONTRAST AGENTS FOR T2-W IMAGING D. Tsivaka1, E. Zygouri2, V. Tangoulis2, I. Tsougos1 1Medical Physics Laboratory, Medical School, University of Thessaly, Larisa, Greece, 2Laboratory of Inorganic Chemistry, Department of Chemistry, University of Patras, Patras, Greece Background: Iron oxide nanoparticles (IONPs) are commonly employed as contrast agents for T2-weighted Magnetic Resonance Imaging (MRI). This study aimed to investigate the effectiveness of H-ScFe and H-GaFe nanoparticles as prospective T2-MRI contrast agents. Materials and Methods: Contrast media were synthesized using H-ScFe and H-GaFe nanoparticles coated with an amorphous silica shell. Four solutions (NPs1, NPs2, NPs3 based on H-GaFe, and NPs4 based on H-ScFe) were prepared using distilled water as solvent. All scans were conducted on a 3.0 T MRI scanner with a custombuilt phantom. T2 relaxation curves were acquired using a singleecho spin-echo sequence with repetition time (TR)= 1500ms, slice thickness= 5mm, flip angle= 90°, and matrix size= 512×512. Echo time (TE) ranged from 15ms to 300ms. T2 relaxation times and R2 relaxation rates were determined using a semi-automated, customdesigned MATLAB code to enhance measurement reproducibility. Transverse relaxivity r2 was defined as the slope of the linear regression generated from plotting the measured relaxation rates against the concentration (C) of the solutions (C=1.8mM for all samples). Results: r2 relaxivity values were calculated 42.4, 34.2, and 56.1 s–1 mM–1 for NPs1, NPs2, and NPs3, respectively, and 29.15 s–1 mM–1 for NPs4. For all solutions, r2 relaxivity values were within the range of commercially available T2 MRI contrast agents (>3 s–1 mM–1 for water solutions). Conclusion: MRI contrast agents based on H-ScFe and H-GaFe nanoparticles demonstrate sufficient performance to enhance contrast in T2-weighted images, suggesting further investigation into the use of IONPs in Magnetic Resonance Imaging. Physica Medica 127S1 (2024) 104615 https://doi.org/10.1016/j.ejmp.2024.104615