Olberg, Sven

Abstract Background Cone‐beam computed tomography (CBCT) is essential for image‐guided particle therapy (IGPT), providing daily patient positioning, anatomical monitoring, and treatment verification. However, conventional single‐detector CBCT suffers from poor soft‐tissue contrast, long acquisition times, and motion artifacts, reducing its effectiveness for adaptive radiotherapy workflows. Addressing these challenges requires an advanced CBCT reconstruction approach capable of enhancing image quality while reducing scan time and radiation exposure. Purpose This study introduces a Dual Detector Combined Scanning (DDCS) CBCT reconstruction algorithm to overcome conventional CBCT limitations. By integrating a dual‐source, orthogonal imaging setup, DDCS significantly reduces scanning time and imaging artifacts. A modified Parker‐weighting algorithm further improves image reconstruction accuracy, ensuring high‐fidelity visualization. The goal is to enhance adaptive radiotherapy by improving real‐time patient positioning, dose verification, and motion management. Methods A dual‐detector CBCT system was designed to acquire orthogonal projections simultaneously, reducing scan time and improving data completeness. A modified Parker‐weighting algorithm corrected angular overlap distortions, ensuring better reconstruction accuracy. The system was evaluated using numerical and physical phantom studies to assess spatial resolution, contrast‐to‐noise ratio (CNR), and artifact reduction. Additionally, a dynamic phantom simulated respiratory motion, validating motion robustness for adaptive radiotherapy. Performance was compared against single‐detector CBCT, focusing on image fidelity, noise suppression, and computational efficiency. Results DDCS‐CBCT demonstrated higher CNR, reduced motion artifacts, and improved spatial resolution, leading to more accurate anatomical visualization. The system's efficiency enables faster, more reliable patient setup in IGPT while maintaining a lower imaging dose. Conclusion The proposed DDCS‐CBCT approach significantly improves imaging accuracy, reduces scan time, and enhances real‐time volumetric guidance in IGRT/IGPT. These findings support the clinical feasibility of the DDCS system and its integration into online adaptive radiotherapy workflows.

AbstractBackgroundAdaptive radiotherapy (ART) workflows have been increasingly adopted to achieve dose escalation and tissue sparing under shifting anatomic conditions, but the necessity of recontouring and the associated time burden hinders a real‐time or online ART workflow. In response to this challenge, approaches to auto‐segmentation involving deformable image registration, atlas‐based segmentation, and deep learning‐based segmentation (DLS) have been developed. Despite the particular promise shown by DLS methods, implementing these approaches in a clinical setting remains a challenge, namely due to the difficulty of curating a data set of sufficient size and quality so as to achieve generalizability in a trained model.PurposeTo address this challenge, we have developed an intentional deep overfit learning (IDOL) framework tailored to the auto‐segmentation task. However, certain limitations were identified, particularly the insufficiency of the personalized dataset to effectively overfit the model. In this study, we introduce a personalized hyperspace learning (PHL)‐IDOL segmentation framework capable of generating datasets that induce the model to overfit specific patient characteristics for medical image segmentation.MethodsThe PHL‐IDOL model is trained in two stages. In the first, a conventional, general model is trained with a diverse set of patient data (n = 100 patients) consisting of CT images and clinical contours. Following this, the general model is tuned with a data set consisting of two components: (a) selection of a subset of the patient data (m < n) using the similarity metrics (mean square error (MSE), peak signal‐to‐noise ratio (PSNR), structural similarity index (SSIM), and the universal quality image index (UQI) values); (b) adjust the CT and the clinical contours using a deformed vector generated from the reference patient and the selected patients using (a). After training, the general model, the continual model, the conventional IDOL model, and the proposed PHL‐IDOL model were evaluated using the volumetric dice similarity coefficient (VDSC) and the Hausdorff distance 95% (HD95%) computed for 18 structures in 20 test patients.ResultsImplementing the PHL‐IDOL framework resulted in improved segmentation performance for each patient. The Dice scores increased from 0.810.05 with the general model, 0.83 for the continual model, 0.83 for the conventional IDOL model to an average of 0.87 with the PHL‐IDOL model. Similarly, the Hausdorff distance decreased from 3.06 with the general model, 2.84 for the continual model, 2.79 for the conventional IDOL model and 2.36 for the PHL‐IDOL model. All the standard deviations were decreased by nearly half of the values comparing the general model and the PHL‐IDOL model.ConclusionThe PHL‐IDOL framework applied to the auto‐segmentation task achieves improved performance compared to the general DLS approach, demonstrating the promise of leveraging patient‐specific prior information in a task central to online ART workflows.

AbstractBackgroundRecent advancements in Deep Learning (DL) methodologies have led to state‐of‐the‐art performance in a wide range of applications especially in object recognition, classification, and segmentation of medical images. However, training modern DL models requires a large amount of computation and long training times due to the complex nature of network structures and the large number of training datasets involved. Moreover, it is an intensive, repetitive manual process to select the optimized configuration of hyperparameters for a given DL network.PurposeIn this study, we present a novel approach to accelerate the training time of DL models via the progressive feeding of training datasets based on similarity measures for medical image segmentation. We term this approach Progressive Deep Learning (PDL).MethodsThe two‐stage PDL approach was tested on the auto‐segmentation task for two imaging modalities: CT and MRI. The training datasets were ranked according to similarity measures between each sample based on Mean Square Error (MSE), Peak Signal‐to‐Noise Ratio (PSNR), Structural Similarity Index (SSIM), and the Universal Quality Image Index (UQI) values. At the start of the training process, a relatively coarse sampling of training datasets with higher ranks was used to optimize the hyperparameters of the DL network. Following this, the samples with higher ranks were used in step 1 to yield accelerated loss minimization in early training epochs and the total dataset was added in step 2 for the remainder of training.ResultsOur results demonstrate that the PDL approach can reduce the training time by nearly half (∼49%) and can predict segmentations (CT U‐net/DenseNet dice coefficient: 0.9506/0.9508, MR U‐net/DenseNet dice coefficient: 0.9508/0.9510) without major statistical difference (Wilcoxon signed‐rank test) compared to the conventional DL approach. The total training times with a fixed cutoff at 0.95 DSC for the CT dataset using DenseNet and U‐Net architectures, respectively, were 17 h, 20 min and 4 h, 45 min in the conventional case compared to 8 h, 45 min and 2 h, 20 min with PDL. For the MRI dataset, the total training times using the same architectures were 2 h, 54 min and 52 min in the conventional case and 1 h, 14 min and 25 min with PDL.ConclusionThe proposed PDL training approach offers the ability to substantially reduce the training time for medical image segmentation while maintaining the performance achieved in the conventional case.

AbstractBackgroundThe growing adoption of magnetic resonance imaging (MRI)‐guided radiation therapy (RT) platforms and a focus on MRI‐only RT workflows have brought the technical challenge of synthetic computed tomography (sCT) reconstruction to the forefront. Unpaired‐data deep learning‐based approaches to the problem offer the attractive characteristic of not requiring paired training data, but the gap between paired‐ and unpaired‐data results can be limiting.PurposeWe present two distinct approaches aimed at improving unpaired‐data sCT reconstruction results: a cascade ensemble that combines multiple models and a personalized training strategy originally designed for the paired‐data setting.MethodsComparisons are made between the following models: (1) the paired‐data fully convolutional DenseNet (FCDN), (2) the FCDN with the Intentional Deep Overfit Learning (IDOL) personalized training strategy, (3) the unpaired‐data CycleGAN, (4) the CycleGAN with the IDOL training strategy, and (5) the CycleGAN as an intermediate model in a cascade ensemble approach. Evaluation of the various models over 25 total patients is carried out using a five‐fold cross‐validation scheme, with the patient‐specific IDOL models being trained for the five patients of fold 3, chosen at random.ResultsIn both the paired‐ and unpaired‐data settings, adopting the IDOL training strategy led to improvements in the mean absolute error (MAE) between true CT images and sCT outputs within the body contour (mean improvement, paired‐ and unpaired‐data approaches, respectively: 38%, 9%) and in regions of bone (52%, 5%), the peak signal‐to‐noise ratio (PSNR; 15%, 7%), and the structural similarity index (SSIM; 6%, <1%). The ensemble approach offered additional benefits over the IDOL approach in all three metrics (mean improvement over unpaired‐data approach in fold 3; MAE: 20%; bone MAE: 16%; PSNR: 10%; SSIM: 2%), and differences in body MAE between the ensemble approach and the paired‐data approach are statistically insignificant.ConclusionsWe have demonstrated that both a cascade ensemble approach and a personalized training strategy designed initially for the paired‐data setting offer significant improvements in image quality metrics for the unpaired‐data sCT reconstruction task. Closing the gap between paired‐ and unpaired‐data approaches is a step toward fully enabling these powerful and attractive unpaired‐data frameworks.