Search

Radioactive ion beams (RIB) are a key focus of current research in nuclear physics. Already long ago it was proposed that they could have applications in cancer therapy. In fact, while charged particle therapy is potentially the most effective radiotherapy technique available, it is highly susceptible to uncertainties in the beam range. RIB are well-suited for image-guided particle therapy, as isotopes that undergo \b{eta}+-decay can be precisely visualized using positron emission tomography (PET), enabling accurate real-time monitoring of the beam range. We successfully treated a mouse osteosarcoma using a radioactive 11C-ion beam. The tumor was located in the neck, in close proximity to the spinal cord, increasing the risk of radiation-induced myelopathy from even slight variations in the beam range caused by anatomical changes or incorrect calibration of the planning CT. We managed to completely control the tumor with the highest dose while minimizing toxicity. Low-grade neurological side effects were correlated to the positron activity measured in the spine. The biological washout of the activity from the tumor volume was dependent on the dose, indicating a potential component of vascular damage at high doses. This experiment marks the first instance of tumor treatment using RIB and paves the way for future clinical applications., Comment: 56 pages, 13 figures, supplements. Video supplements available on request

Diabetic retinopathy (DR) is a consequence of diabetes mellitus characterized by vascular damage within the retinal tissue. Timely detection is paramount to mitigate the risk of vision loss. However, training robust grading models is hindered by a shortage of annotated data, particularly for severe cases. This paper proposes a framework for controllably generating high-fidelity and diverse DR fundus images, thereby improving classifier performance in DR grading and detection. We achieve comprehensive control over DR severity and visual features (optic disc, vessel structure, lesion areas) within generated images solely through a conditional StyleGAN, eliminating the need for feature masks or auxiliary networks. Specifically, leveraging the SeFa algorithm to identify meaningful semantics within the latent space, we manipulate the DR images generated conditionally on grades, further enhancing the dataset diversity. Additionally, we propose a novel, effective SeFa-based data augmentation strategy, helping the classifier focus on discriminative regions while ignoring redundant features. Using this approach, a ResNet50 model trained for DR detection achieves 98.09% accuracy, 99.44% specificity, 99.45% precision, and an F1-score of 98.09%. Moreover, incorporating synthetic images generated by conditional StyleGAN into ResNet50 training for DR grading yields 83.33% accuracy, a quadratic kappa score of 87.64%, 95.67% specificity, and 72.24% precision. Extensive experiments conducted on the APTOS 2019 dataset demonstrate the exceptional realism of the generated images and the superior performance of our classifier compared to recent studies., Comment: 30 pages, 17 figures