29 June 2023 in Cancer Research Communications
Artificial intelligence (AI) and machine learning (ML) are becoming critical in developing and deploying personalized medicine and targeted clinical trials. Recent advances in ML have enabled the integration of wider ranges of data including both medical records and imaging (radiomics). However, the development of prognostic models is complex as no modeling strategy is universally superior to others and validation of developed models requires large and diverse datasets to demonstrate that prognostic models developed (regardless of method) from one dataset are applicable to other datasets both internally and externally. Using a retrospective dataset of 2,552 patients from a single institution and a strict evaluation framework that included external validation on three external patient cohorts (873 patients), we crowdsourced the development of ML models to predict overall survival in head and neck cancer (HNC) using electronic medical records (EMR) and pretreatment radiological images. To assess the relative contributions of radiomics in predicting HNC prognosis, we compared 12 different models using imaging and/or EMR data. The model with the highest accuracy used multitask learning on clinical data and tumor volume, achieving high prognostic accuracy for 2-year and lifetime survival prediction, outperforming models relying on clinical data only, engineered radiomics, or complex deep neural network architecture. However, when we attempted to extend the best performing models from this large training dataset to other institutions, we observed significant reductions in the performance of the model in those datasets, highlighting the importance of detailed population-based reporting for AI/ML model utility and stronger validation frameworks. We have developed highly prognostic models for overall survival in HNC using EMRs and pretreatment radiological images based on a large, retrospective dataset of 2,552 patients from our institution.Diverse ML approaches were used by independent investigators. The model with the highest accuracy used multitask learning on clinical data and tumor volume.External validation of the top three performing models on three datasets (873 patients) with significant differences in the distributions of clinical and demographic variables demonstrated significant decreases in model performance. Significance: ML combined with simple prognostic factors outperformed multiple advanced CT radiomics and deep learning methods. ML models provided diverse solutions for prognosis of patients with HNC but their prognostic value is affected by differences in patient populations and require extensive validation.
1 April 2016 in British Journal of Radiology
Objective:The aim of this study was to determine which physical delivery parameter changes are most suitable for multiple-level dose-painting treatment plans with helical tomotherapy (HT).Methods:A total of 96 treatment plans were generated for 12 patients who had undergone fluorine-18 fludeoxyglucose positron emission tomography/CT (18F-FDG-PET/CT) scan to plan head and neck cancer treatment. Based on these PET-CT images, the dose was escalated to 96 Gy in 32 fractions as a function of PET intensity values. The intensity-based prescription was converted into seven discrete dose levels. For the same patient, different HT plans were optimized by varying parameters such as field width (FW), pitch (PF) and modulation factor (MF). Dose conformity was evaluated using quality–volume histograms, quality factors (QFs), weighted index of achievement (IOAw), coldness (IOCw) and hotness (IOHw). Moreover, doses to organs at risk (OARs), target volumes and execution time were analyzed.Results:Median QFs were the best ...
21 April 2023 in Zeitschrift für Medizinische Physik
The shape of the energy spectrum is an essential component of any electron beam Monte Carlo model. Due to specialized equipment and the long measurement time for the direct methods for determining the energy spectrum, attractive alternatives are backward spectrum reconstructions from the measured data. One such approach is solving the first-degree Fredholm integral equation with appropriate regularization. It makes it possible to calculate the depth distribution as the sum of the distributions from monoenergetic beams. This study aims to determine the optimal value of the regularization parameter for the problem of determining the spectrum of the electron beam produced by a mobile accelerator used during intraoperative radiotherapy. The Geant4 package was used to generate the distributions of deep doses for monoenergetic beams for two models with different degrees of complexity, i.e. simple (theoretical) and full (for the mobile accelerator). The dose distributions for four different shapes of energy spectrum (for each model) were obtained similarly. They were established as the reference data for further calculations. The Dual Annealing optimization method was used to obtain the reconstructed spectrum. The multiple optimizations that differ by the regularization parameter (ranging from 0 to 1) were performed. For each reconstruction, similarity indicators of the energy spectrum and the dose distribution to the referenced data were calculated to determine the optimal regularization parameters. Optimal regularization parameters determined by similarity indicators for the spectrum and the dose distribution differ for geometry models considered in the study. The regularization parameter for the simple geometry ranged from 0.03 to 0.05, while for full geometry, they were from 0.05 to 0.06. The results for conventional linear accelerators found in the literature range from 0.5 to 1.1. The Dual Annealing optimization method can be effectively used to solve the Fredholm equation with Tikhonov regularization to reconstruct an electron beam's energy spectrum. The regularization parameter value depends on the beam-forming system. Its value for the mobile accelerator considered in the study ranges from 0.05 to 0.06, depending on the nominal beam energy value.
23 June 2017 in British Journal of Radiology
Objective: To determine and quantify the percentage dose increase to organs at risk (OARs) with multiple-level dose painting (DP) for patients with head and neck cancer in comparison with standard regimen. Methods: 12 patients who had undergone fluorine-18 fludeoxyglucose (18F-FDG) positron emission tomography (PET)/CT scan were retrospectively enrolled. Two treatment plans—one using DP escalation and one without—were optimized for each patient base on PET/CT data. The following variables were assessed: dose to OARs and target volumes; execution time; equivalent uniform dose; and normal tissue complication probability. Results: No statistically significant differences in beam-on time were observed between plans with and without DP. However, significantly higher doses were observed for all DP-escalated plans in the OARs, with only two exceptions: the brain stem and V 60Gy for the mandible. Multiple-level DP resulted in dose increases ranging from 3.0% to 12.9%, depending on the OAR. The largest increase was seen for the parotid glands and the smallest for the mandible. Significant differences in the equivalent uniform dose were observed only for the parotid glands and spinal column, where the dose without DP was lower. The normal tissue complication probability for most OARs was very small. Conclusion: Importantly, even though DP escalation resulted in higher doses to OARs vs conventional treatment planning, these usually did not exceed the dose tolerance levels. However, clinical trials are necessary to confirm the benefits of DP and to guarantee no additional toxicity. Advances in knowledge: Multiple-level DP by numbers resulted in 3.0–12.9% dose increase, depending on the OAR. Our findings may suggest that DP escalation to very high doses is feasible for about 83% of patients without higher toxicity; however, it still should be confirmed on a larger group of patients.