Principles of artificial intelligence in radiooncology.

PURPOSE: In the rapidly expanding field of artificial intelligence (AI) there is a wealth of literature detailing the myriad applications of AI, particularly in the realm of deep learning. However, a review that elucidates the technical principles of deep learning as relevant to radiation oncology in an easily understandable manner is still notably lacking. This paper aims to fill this gap by providing a comprehensive guide to the principles of deep learning that is specifically tailored toward radiation oncology. METHODS: In light of the extensive variety of AI methodologies, this review selectively concentrates on the specific domain of deep learning. It emphasizes the principal categories of deep learning models and delineates the methodologies for training these models effectively. RESULTS: This review initially delineates the distinctions between AI and deep learning as well as between supervised and unsupervised learning. Subsequently, it elucidates the fundamental principles of major deep learning models, encompassing multilayer perceptrons (MLPs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), transformers, generative adversarial networks (GANs), diffusion-based generative models, and reinforcement learning. For each category, it presents representative networks alongside their specific applications in radiation oncology. Moreover, the review outlines critical factors essential for training deep learning models, such as data preprocessing, loss functions, optimizers, and other pivotal training parameters including learning rate and batch size. CONCLUSION: This review provides a comprehensive overview of deep learning principles tailored toward radiation oncology. It aims to enhance the understanding of AI-based research and software applications, thereby bridging the gap between complex technological concepts and clinical practice in radiation oncology.

Comprehensive multimodal deep learning survival prediction enabled by a transformer architecture: A multicenter study in glioblastoma.

Background: This research aims to improve glioblastoma survival prediction by integrating MR images, clinical, and molecular-pathologic data in a transformer-based deep learning model, addressing data heterogeneity and performance generalizability. Methods: We propose and evaluate a transformer-based nonlinear and nonproportional survival prediction model. The model employs self-supervised learning techniques to effectively encode the high-dimensional MRI input for integration with nonimaging data using cross-attention. To demonstrate model generalizability, the model is assessed with the time-dependent concordance index (Cdt) in 2 training setups using 3 independent public test sets: UPenn-GBM, UCSF-PDGM, and Rio Hortega University Hospital (RHUH)-GBM, each comprising 378, 366, and 36 cases, respectively. Results: The proposed transformer model achieved a promising performance for imaging as well as nonimaging data, effectively integrating both modalities for enhanced performance (UCSF-PDGM test-set, imaging Cdt 0.578, multimodal Cdt 0.672) while outperforming state-of-the-art late-fusion 3D-CNN-based models. Consistent performance was observed across the 3 independent multicenter test sets with Cdt values of 0.707 (UPenn-GBM, internal test set), 0.672 (UCSF-PDGM, first external test set), and 0.618 (RHUH-GBM, second external test set). The model achieved significant discrimination between patients with favorable and unfavorable survival for all 3 datasets (log-rank P 1.9 × 10-8, 9.7 × 10-3, and 1.2 × 10-2). Comparable results were obtained in the second setup using UCSF-PDGM for training/internal testing and UPenn-GBM and RHUH-GBM for external testing (Cdt 0.670, 0.638, and 0.621). Conclusions: The proposed transformer-based survival prediction model integrates complementary information from diverse input modalities, contributing to improved glioblastoma survival prediction compared to state-of-the-art methods. Consistent performance was observed across institutions supporting model generalizability.

Outcomes of I-125 Low-Dose-Rate Brachytherapy in Patients with Localized Prostate Cancer: A Comprehensive Analysis from a Specialized Tertiary Referral Center.

Low-dose-rate (LDR) brachytherapy with I-125 seeds is one of the most common primary tumor treatments for low-risk and low-intermediate-risk prostate cancer. This report aimed to present an analysis of single-institution long-term results. We analyzed the treatment outcomes of 119 patients with low- and intermediate-risk prostate cancer treated with LDR brachytherapy at our institution between 2014 and 2020. The analysis focused on biochemical recurrence rates (BRFS), overall survival (OS), cumulative local recurrence rate (CLRR), and the incidence of acute and late toxicities. Patient-reported quality of life measures were also evaluated to provide a holistic view on the treatment's impact. The median follow-up period was 46 months. CLRR was 3.3% (4/119), five-year BRFS was 87%, and the five-year OS rate was 95%. Dysuria was the most common acute urinary toxicity, reported in 26.0% of patients as grade 1 and 13.4% as grade 2. As a late side effect, 12.6% of patients experienced mild dysuria. Sexual dysfunction persisted in 6.7% of patients as grade 1, 7.5% as grade 2, and 10.0% as grade 3. LDR brachytherapy in patients with prostate cancer is an effective treatment, with favorable clinical outcomes and manageable toxicity. The low CLRR and high OS rates, as well as low incidence of severe side effects, support the continued use of LDR brachytherapy as a primary treatment modality for localized prostate cancer.

Treatment with (chemo)-radiation in old patients (≥76 years of age) with newly diagnosed non-metastatic squamous cell cancer of the head and neck region: real-world data from a tertiary referral center.

Purpose: Treatment of patients with cancer of the head and neck region is in focus in a multitude of studies. Of these patients, one patient group, those aged 76 and more, is mostly underrepresented despite requiring thorough and well-reasoned treatment decisions to offer curative treatment. This study investigates real-world data on curative treatment of old (≥76 years) patients with newly diagnosed squamous cell carcinoma of the head and neck region (HNSCC). Patients and methods: Between January 2010 and December 2021, we identified 71 patients older than 76 years with newly diagnosed HNSCC and cM0 at the Department of Radiation Oncology of the University Hospital of Erlangen-Nuremberg. Using electronic medical records, we analyzed treatment patterns and outcomes in terms of overall survival (OS), progression-free survival (PFS), and locoregional control (LRC) rate. Additionally, we performed univariate risk analysis and Cox regression in order to identify predictive factors associated with the abovementioned treatment outcomes. Results: The median follow-up was 18 months. OS was 83%, 79%, and 72% after 1 year, 2 years, and 3 years, respectively. PFS was 69%, 54%, and 46% after 1 year, 2 years, and 3 years, respectively. A total of 34 (48%) patients were treated with standard therapy according to current guidelines. The reasons for deviation from standard therapy before or during treatment were as follows: unfitness for cisplatin-based chemotherapy (n = 37), reduction of chemotherapy (n = 3), and dose reduction/interruption of radiotherapy (n = 8). Carboplatin-based systemic therapy showed improved PFS compared to cisplatin or cetuximab (60 vs. 28 vs. 15 months, p = 0.037) but without impact on OS (83 vs. 52 vs. 38 months, p = 0.807). Oropharyngeal tumor localization (p = 0.026) and combined treatment (surgery and postoperative treatment) (p = 0.008) were significant predictors for a better OS. In multivariate analysis, oropharyngeal tumor localization (p = 0.011) and combined treatment (p = 0.041) showed significantly increased PFS. After 1 year, 2 years, and 3 years, the cumulative incidence of locoregional recurrences (LRRs) was 13%, 24%, and 27%, respectively, and was significantly decreased in patients with oropharyngeal tumor localization (p = 0.037). Conclusions: Adherence to treatment protocols for radiotherapy alone in old patients with HNSCC is good, whereas the application of concurrent chemotherapy often deviates from guidelines in terms of de-escalation. An important risk factor for decreased OS, PFS, and a higher rate of LRR appears to be non-oropharyngeal tumor location in old patients.

"sCT-Feasibility" - a feasibility study for deep learning-based MRI-only brain radiotherapy.

BACKGROUND: Radiotherapy (RT) is an important treatment modality for patients with brain malignancies. Traditionally, computed tomography (CT) images are used for RT treatment planning whereas magnetic resonance imaging (MRI) images are used for tumor delineation. Therefore, MRI and CT need to be registered, which is an error prone process. The purpose of this clinical study is to investigate the clinical feasibility of a deep learning-based MRI-only workflow for brain radiotherapy, that eliminates the registration uncertainty through calculation of a synthetic CT (sCT) from MRI data. METHODS: A total of 54 patients with an indication for radiation treatment of the brain and stereotactic mask immobilization will be recruited. All study patients will receive standard therapy and imaging including both CT and MRI. All patients will receive dedicated RT-MRI scans in treatment position. An sCT will be reconstructed from an acquired MRI DIXON-sequence using a commercially available deep learning solution on which subsequent radiotherapy planning will be performed. Through multiple quality assurance (QA) measures and reviews during the course of the study, the feasibility of an MRI-only workflow and comparative parameters between sCT and standard CT workflow will be investigated holistically. These QA measures include feasibility and quality of image guidance (IGRT) at the linear accelerator using sCT derived digitally reconstructed radiographs in addition to potential dosimetric deviations between the CT and sCT plan. The aim of this clinical study is to establish a brain MRI-only workflow as well as to identify risks and QA mechanisms to ensure a safe integration of deep learning-based sCT into radiotherapy planning and delivery. DISCUSSION: Compared to CT, MRI offers a superior soft tissue contrast without additional radiation dose to the patients. However, up to now, even though the dosimetrical equivalence of CT and sCT has been shown in several retrospective studies, MRI-only workflows have still not been widely adopted. The present study aims to determine feasibility and safety of deep learning-based MRI-only radiotherapy in a holistic manner incorporating the whole radiotherapy workflow. TRIAL REGISTRATION: NCT06106997.

Deep Learning and Registration-Based Mapping for Analyzing the Distribution of Nodal Metastases in Head and Neck Cancer Cohorts: Informing Optimal Radiotherapy Target Volume Design.

We introduce a deep-learning- and a registration-based method for automatically analyzing the spatial distribution of nodal metastases (LNs) in head and neck (H/N) cancer cohorts to inform radiotherapy (RT) target volume design. The two methods are evaluated in a cohort of 193 H/N patients/planning CTs with a total of 449 LNs. In the deep learning method, a previously developed nnU-Net 3D/2D ensemble model is used to autosegment 20 H/N levels, with each LN subsequently being algorithmically assigned to the closest-level autosegmentation. In the nonrigid-registration-based mapping method, LNs are mapped into a calculated template CT representing the cohort-average patient anatomy, and kernel density estimation is employed to estimate the underlying average 3D-LN probability distribution allowing for analysis and visualization without prespecified level definitions. Multireader assessment by three radio-oncologists with majority voting was used to evaluate the deep learning method and obtain the ground-truth distribution. For the mapping technique, the proportion of LNs predicted by the 3D probability distribution for each level was calculated and compared to the deep learning and ground-truth distributions. As determined by a multireader review with majority voting, the deep learning method correctly categorized all 449 LNs to their respective levels. Level 2 showed the highest LN involvement (59.0%). The level involvement predicted by the mapping technique was consistent with the ground-truth distribution (p for difference 0.915). Application of the proposed methods to multicenter cohorts with selected H/N tumor subtypes for informing optimal RT target volume design is promising.

Chemoradiotherapy plus hyperthermia (CRTH) versus chemoradiotherapy (CRT) alone in neoadjuvant treatment of soft tissue sarcoma: tumor response, treatment toxicity and disease control.

INTRODUCTION: Neoadjuvant chemotherapy and radiotherapy for the management of soft tissue sarcomas (STS) are still preferably delivered sequentially, with or without concurrent hyperthermia. Concurrent delivery of chemo-, radio- and thermotherapy may produce synergistic effects and reduce chemotherapy-free intervals. The few available studies suggest that concurrent chemoradiation (CRT) has a greater local effect. Data on the efficacy and toxicity of adding hyperthermia to CRT (CRTH) are sparse. MATERIALS AND METHODS: A cohort of 101 patients with STS of the extremities and trunk who received CRT (n = 33) or CRTH (n = 68) before resection of macroscopic tumor (CRT: n = 19, CRTH: n = 49) or re-resection following a non-oncological resection, so called 'whoops procedure', (CRT: n = 14, CRTH: n = 19) were included in this retrospective study. CRT consisted of two cycles of doxorubicine (50 mg/m2 on d2) plus ifosfamide (1500 mg/m2 on d1-5, q28) plus radiation doses of up to 60 Gy. Hyperthermia was delivered in two sessions per week. RESULTS: All patients received the minimum dose of 50 Gy. Median doses of ifosfamide and doxorubicin were comparable between CRT (75%/95%) and CRTH (78%/97%). The median number of hyperthermia sessions was seven. There were no differences in acute toxicities. Major wound complications occurred in 15% (CRT) vs. 25% (CRTH) (p = 0.19). In patients with macroscopic disease, the addition of hyperthermia resulted in a tendency toward improved remission: regression ≥90% occurred in 21/48 (CRTH) vs. 4/18 (CRT) patients (p = 0.197). With a median postoperative follow-up of 72 months, 6-year local control and overall survival rates for CRTH vs. CRT alone were 85 vs. 78% (p = 0.938) and 79 vs. 71% (p = 0.215). CONCLUSIONS: Both CRT and CRTH are well tolerated with an expected rate of wound complications. The results suggest that adding hyperthermia may improve tumor response.

Low-dose radiotherapy of osteoarthritis: from biological findings to clinical effects-challenges for future studies.

Osteoarthritis (OA) is one of the most common and socioeconomically relevant diseases, with rising incidence and prevalence especially with regard to an ageing population in the Western world. Over the decades, the scientific perception of OA has shifted from a simple degeneration of cartilage and bone to a multifactorial disease involving various cell types and immunomodulatory factors. Despite a wide range of conventional treatment modalities available, a significant proportion of patients remain treatment refractory. Low-dose radiotherapy (LDRT) has been used for decades in the treatment of patients with inflammatory and/or degenerative diseases and has proven a viable option even in cohorts of patients with a rather poor prognosis. While its justification mainly derives from a vast body of empirical evidence, prospective randomized trials have until now failed to prove the effectiveness of LDRT. Nevertheless, over the decades, adaptions of LDRT treatment modalities have evolved using lower dosages with establishment of different treatment schedules for which definitive clinical proof is still pending. Preclinical research has revealed that the immune system is modulated by LDRT and very recently osteoimmunological mechanisms have been described. Future studies and investigations further elucidating the underlying mechanisms are an essential key to clarify the optimal patient stratification and treatment procedure, considering the patients' inflammatory status, age, and sex. The present review aims not only to present clinical and preclinical knowledge about the mechanistic and beneficial effects of LDRT, but also to emphasize topics that will need to be addressed in future studies. Further, a concise overview of the current status of the underlying radiobiological knowledge of LDRT for clinicians is given, while seeking to stimulate further translational research.

Oligometastatic head and neck cancer: Which patients benefit from radical local treatment of all tumour sites?

BACKGROUND: There is a large lack of evidence for optimal treatment in oligometastatic head and neck cancer and it is especially unclear which patients benefit from radical local treatment of all tumour sites. METHODS: 40 patients with newly diagnosed oligometastatic head and neck cancer received radical local treatment of all tumour sites from 14.02.2008 to 24.08.2018. Primary endpoint was overall survival. Time to occurrence of new distant metastases and local control were evaluated as secondary endpoints as well as prognostic factors in univariate und multivariate Cox's regression analysis. To investigate the impact of total tumour volume on survival, all tumour sites were segmented on baseline imaging. RESULTS: Radical local treatment included radiotherapy in 90% of patients, surgery in 25% and radiofrequency ablation in 3%. Median overall survival from first diagnosis of oligometastatic disease was 23.0 months, 2-year survival was 48%, 3-year survival was 37%, 4-year survival was 24% and 5-year survival was 16%. Median time to occurrence of new distant metastases was 11.6 months with freedom from new metastases showing a tail pattern after 3 years of follow-up (22% at 3, 4- and 5-years post-treatment). In multivariate analysis, better ECOG status, absence of bone and brain metastases and lower total tumour volume were significantly associated with improved survival, whereas the number of metastases and involved organ sites was not. CONCLUSIONS: Radical local treatment in oligometastatic head and neck cancer shows promising outcomes and needs to be further pursued. Patients with good performance status, absence of brain and bone metastases and low total tumour volume were identified as optimal candidates for radical local treatment in oligometastatic head and neck cancer and should be considered for selection in future prospective trials.

Protein kinase A regulates inflammatory pain sensitization by modulating HCN2 channel activity in nociceptive sensory neurons.

Several studies implicated cyclic adenosine monophosphate (cAMP) as an important second messenger for regulating nociceptor sensitization, but downstream targets of this signaling pathway which contribute to neuronal plasticity are not well understood. We used a Cre/loxP-based strategy to disable the function of either HCN2 or PKA selectively in a subset of peripheral nociceptive neurons and analyzed the nociceptive responses in both transgenic lines. A near-complete lack of sensitization was observed in both mutant strains when peripheral inflammation was induced by an intradermal injection of 8br-cAMP. The lack of HCN2 as well as the inhibition of PKA eliminated the cAMP-mediated increase of calcium transients in dorsal root ganglion neurons. Facilitation of Ih via cAMP, a hallmark of the Ih current, was abolished in neurons without PKA activity. Collectively, these results show a significant contribution of both genes to inflammatory pain and suggest that PKA-dependent activation of HCN2 underlies cAMP-triggered neuronal sensitization.