RESUMEN
In interventional radiology patient care can be improved by accurately assessing peak skin dose (PSD) from procedures, as it is the main predictor for tissue-reactions such as erythema. Historically, high skin dose procedures performed in radiology departments were almost exclusively planar fluoroscopy. However, with the increase in use of technologies involving repeated or adjacent computed tomography (CT) such as CT fluoroscopy and multi-modality rooms, the peak skin dose delivered by CT needs to be considered. In this paper, a model to estimate the PSD delivered to a patient undergoing CT has been developed to assist in determining the overall PSD. This model relates the PSD to the device-reported CT Dose Index (CTDIvol) by accounting for a variety of CT technique and patient factors. It includes a novel method for estimating dose contributions as a function of patient or phantom size, scanner geometry, and physical measurement of lateral and depth-based beam profiles. Physical measurements of PSD using radiochromic film on several phantoms have been used to determine needed model parameters. The resulting fitted model was found to agree with measured data to a standard deviation of 5.1% for the data used to fit the model, and 6.8% for measurements that were not used for fitting the model. Two methods for adapting the model for specific scanners are provided, one based on local PSD measurements with radiochromic film and another using CTDIvol measurements. The model, when suitably adapted, can accurately assess individual patients' CT PSD. This information can be integrated with radiation exposure data from other modalities, such as planar fluoroscopy, to predict the overall risk of tissue reactions, allowing for more tailored patient care.
Asunto(s)
Fantasmas de Imagen , Dosis de Radiación , Piel , Tomografía Computarizada por Rayos X , Humanos , Piel/diagnóstico por imagen , Relación Dosis-Respuesta en la RadiaciónRESUMEN
Dosimetry audits for passive motion management require dynamically-acquired measurements in a moving phantom to be compared to statically calculated planned doses. This study aimed to characterise the relationship between planning and delivery errors, and the measured dose in the Imaging and Radiation Oncology Core (IROC) thorax phantom, to assess different audit scoring approaches. Treatment plans were created using a 4DCT scan of the IROC phantom, equipped with film and thermoluminescent dosimeters (TLDs). Plans were created on the average intensity projection from all bins. Three levels of aperture complexity were explored: dynamic conformal arcs (DCAT), low-, and high-complexity volumetric modulated arcs (VMATLo, VMATHi). Simulated-measured doses were generated by modelling motion using isocenter shifts. Various errors were introduced including incorrect setup position and target delineation. Simulated-measured film doses were scored using gamma analysis and compared within specific regions of interest (ROIs) as well as the entire film plane. Positional offsets were estimated based on isodoses on the film planes, and point doses within TLD contours were compared. Motion-induced differences between planned and simulated-measured doses were evident even without introduced errors Gamma passing rates within target-centred ROIs correlated well with error-induced dose differences, while whole film passing rates did not. Isodose-based setup position measurements demonstrated high sensitivity to errors. Simulated point doses at TLD locations yielded erratic responses to introduced errors. ROI gamma analysis demonstrated enhanced sensitivity to simulated errors compared to whole film analysis. Gamma results may be further contextualized by other metrics such as setup position or maximum gamma.
Asunto(s)
Movimiento , Fantasmas de Imagen , Planificación de la Radioterapia Asistida por Computador , Tórax , Tórax/diagnóstico por imagen , Planificación de la Radioterapia Asistida por Computador/métodos , Humanos , Radiometría/instrumentación , Dosificación Radioterapéutica , Radioterapia de Intensidad Modulada , Tomografía Computarizada Cuatridimensional , Movimiento (Física)RESUMEN
PURPOSE: Artificial intelligence (AI)-based auto-segmentation models hold promise for enhanced efficiency and consistency in organ contouring for adaptive radiation therapy and radiation therapy planning. However, their performance on pediatric computed tomography (CT) data and cross-scanner compatibility remain unclear. This study aimed to evaluate the performance of AI-based auto-segmentation models trained on adult CT data when applied to pediatric data sets and explore the improvement in performance gained by including pediatric training data. It also examined their ability to accurately segment CT data acquired from different scanners. METHODS AND MATERIALS: Using the nnU-Net framework, segmentation models were trained on data sets of adult, pediatric, and combined CT scans for 7 pelvic/thoracic organs. Each model was trained on 290 to 300 cases per category and organ. Training data sets included a combination of clinical data and several open repositories. The study incorporated a database of 459 pediatric (0-16 years) CT scans and 950 adults (>18 years), ensuring all scans had human expert ground-truth contours of the selected organs. Performance was evaluated based on Dice similarity coefficients (DSC) of the model-generated contours. RESULTS: AI models trained exclusively on adult data underperformed on pediatric data, especially for the 0 to 2 age group: mean DSC was below 0.5 for the bladder and spleen. The addition of pediatric training data demonstrated significant improvement for all age groups, achieving a mean DSC of above 0.85 for all organs in every age group. Larger organs like the liver and kidneys maintained consistent performance for all models across age groups. No significant difference emerged in the cross-scanner performance evaluation, suggesting robust cross-scanner generalization. CONCLUSIONS: For optimal segmentation across age groups, it is important to include pediatric data in the training of segmentation models. The successful cross-scanner generalization also supports the real-world clinical applicability of these AI models. This study emphasizes the significance of data set diversity in training robust AI systems for medical image interpretation tasks.