Over the last decade there has been an extensive evolution in the Artificial Intelligence (AI) field. Modern radiation oncology is based on the exploitation of advanced computational methods aiming to personalization and high diagnostic and therapeutic precision. The quantity of the available imaging data and the increased developments of Machine Learning (ML), particularly Deep Learning (DL), triggered the research on uncovering "hidden" biomarkers and quantitative features from anatomical and functional medical images. Deep Neural Networks (DNN) have achieved outstanding performance and broad implementation in image processing tasks. Lately, DNNs have been considered for radiomics and their potentials for explainable AI (XAI) may help classification and prediction in clinical practice. However, most of them are using limited datasets and lack generalized applicability. In this study we review the basics of radiomics feature extraction, DNNs in image analysis, and major interpretability methods that help enable explainable AI. Furthermore, we discuss the crucial requirement of multicenter recruitment of large datasets, increasing the biomarkers variability, so as to establish the potential clinical value of radiomics and the development of robust explainable AI models.

Objective: A methodology is introduced for the development of an internal dosimetry prediction toolkit for nuclear medical pediatric applications. The proposed study exploits Artificial Intelligence techniques using Monte Carlo simulations as ground truth for accurate prediction of absorbed doses per organ prior to the imaging acquisition considering only personalized anatomical characteristics of any new pediatric patient. Approach: GATE Monte Carlo simulations were performed using a population of computational pediatric models to calculate the specific absorbed dose rates (SADRs) in several organs. A simulated dosimetry database was developed for 28 pediatric phantoms (age range 2–17 years old, both genders) and 5 different radiopharmaceuticals. Machine Learning regression models were trained on the produced simulated dataset, with leave one out cross validation for the prediction model evaluation. Hyperparameter optimization and ensemble learning techniques for a variation of input features were applied for achieving the best predictive power, leading to the development of a SADR prediction toolkit for any new pediatric patient for the studied organs and radiopharmaceuticals. Main results . SADR values for 30 organs of interest were calculated via Monte Carlo simulations for 28 pediatric phantoms for the cases of five radiopharmaceuticals. The relative percentage uncertainty in the extracted dose values per organ was lower than 2.7%. An internal dosimetry prediction toolkit which can accurately predict SADRs in 30 organs for five different radiopharmaceuticals, with mean absolute percentage error on the level of 8% was developed, with specific focus on pediatric patients, by using Machine Learning regression algorithms, Single or Multiple organ training and Artificial Intelligence ensemble techniques. Significance: A large simulated dosimetry database was developed and utilized for the training of Machine Learning models. The developed predictive models provide very fast results (90% with respect to the ground truth of Monte Carlo, considering personalized anatomical characteristics and the biodistribution of each radiopharmaceutical. The proposed method is applicable to other medical dosimetry applications in different patients’ populations.

The aim of this study was to evaluate the suitability of three approaches in producing physical breast phantoms using the low-cost 3D printing fused filament fabrication (FFF) technique. Three physical phantoms were produced: the first (PB1) is based on PLA filament extruded with different extrusion rates and constant printing speed; the second (PB2) and third (PB3) were each printed with two filaments: ABS-ASA for PB2 and HIPS-ASA for PB3. The phantoms were scanned using a clinical CT facility at different kilovoltage (kV) settings. Objective analysis comprised power spectrum, fractal dimensions, histogram metrics (skewness, kurtosis), and Hounsfield Unit (HU) values. Experienced radiologists performed a subjective assessment. All phantoms demonstrated β indexes in the range of 1.39-2.72, with PB2 showing the lowest values. The measured HU values for the breast tissues as a whole increased with increasing kV. PB1 showed very good results in terms of measured HU values for glandular tissue, skin, and adipose tissue. In PB2, ABS proved to be well suited for representing adipose tissue, whereas HIPS exhibited lower than expected HU values. Similarly, in PB3, ASA showed HU values that were lower than those reported in the literature. These discrepancies highlight the need for further optimization of printing parameters for HIPS and ASA materials. Subjective evaluation revealed no significant differences in the visual quality of the phantoms. FFF printing challenges hinder widespread anthropomorphic phantom production. Further research is needed for dual-material printing.

In the current work, a pix2pix conditional generative adversarial network has been evaluated as a potential solution for generating adequately accurate synthesized morphological X-ray images by translating standard photographic images of mice. Such an approach will benefit 2D functional molecular imaging techniques, such as planar radioisotope and/or fluorescence/bioluminescence imaging, by providing high-resolution information for anatomical mapping, but not for diagnosis, using conventional photographic sensors. Planar functional imaging offers an efficient alternative to biodistribution ex vivo studies and/or 3D high-end molecular imaging systems since it can be effectively used to track new tracers and study the accumulation from zero point in time post-injection. The superimposition of functional information with an artificially produced X-ray image may enhance overall image information in such systems without added complexity and cost. The network has been trained in 700 input (photography)/ground truth (X-ray) paired mouse images and evaluated using a test dataset composed of 80 photographic images and 80 ground truth X-ray images. Performance metrics such as peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM) and Fréchet inception distance (FID) were used to quantitatively evaluate the proposed approach in the acquired dataset.

Radiomics as a novel quantitative approach to medical imaging is an emerging area in the field of radiology. Artificial Intelligence offers promising tools for exploiting and analyzing radiomics. The objective of the present study is to propose a methodology for the design, development, and evaluation of Machine Learning (ML) models for the prediction of the recovery progress of skeletal muscle injury over time in rats using radiomics. Radiomics were extracted from contrast enhanced Computed Tomography (CT) data and ML algorithms were trained and compared for their predictive value based on different CT imaging parameters. Ten different ML regression algorithms were tested and the optimal combination of radiomics for each algorithm and CT imaging parameter settings combination was studied. The best ensemble learning model, trained on the 70kVp, 100mA imaging parameter dataset, achieved a Mean Absolute Error score of 1.22. The results suggest that radiomics extracted from CT images can be used as input in ML regression algorithms to predict the volume of a skeletal muscle injury in rats. Moreover, the results show that CT imaging settings impact the predictive performance of the ML regression models, indicating that lower values of tube current and peak kilovoltage contribute to more accurate predictions.

Background/Objectives: Lutetium-177 is a widely used radioisotope in targeted radionuclide therapy, particularly for treating certain types of cancers relying on beta and low-energy gamma emissions, making it suitable for both therapeutic and post-therapy monitoring purposes. The purpose of this study was to evaluate the technical parameters for developing a prototype portable gamma camera dedicated to ¹⁷⁷Lu imaging applications. Methods: The well-validated GATE Monte Carlo toolkit was used to study the characteristics of the system and evaluate its performance in terms of spatial resolution, sensitivity, and image quality. For this purpose, a series of Monte Carlo simulations were executed, modeling a channel-edge aperture pinhole collimator incorporating a variety of computational phantoms. The final configuration of the prototype was standardized, incorporating the crystal size, collimator design, shielding, and the optimal FOV. After the development of the actual prototype camera, the system was also validated experimentally on the same setups as the simulations. Results: The final configuration of the prototype imaging system was standardized based on simulation results and then experimentally validated using physical phantoms under equivalent conditions. A minification of 1:5, spatial resolution of 1.0 cm, and sensitivity of 5.2 Cps/MBq at 10 cm distance source-to-collimator distance were assessed and confirmed. The experimental results agreed within 5% of simulated values. Conclusions: This study establishes the technical feasibility and foundational performance of a portable pinhole imaging system for potential clinical use in ¹⁷⁷Lu imaging workflows and thereby improving therapeutic effectiveness.