Tyler James Bradshaw

This study aimed to evaluate the prognostic value of conventional and advanced PET metrics for predicting progression-free survival in high-risk pediatric Hodgkin lymphoma (HL), using data from the Children's Oncology Group AHOD1331 trial. Methods: This retrospective analysis included 558 patients from 150 institutions. Conventional PET metrics and radiomics features were extracted from both baseline and interim-therapy PET images. Clinical variables, including demographic characteristics and common risk factors, were also considered. A standardized outcome-modeling pipeline was developed, comprising feature selection and penalized Cox regression with elastic net regularization (CoxNet). Models were trained and evaluated using nested cross-validation stratified by institution, ensuring consistent external testing. We investigated whether conventional PET metrics added prognostic value beyond clinical variables and whether radiomics or interim PET-derived features offered further benefit. Additionally, we compared the prognostic performance of features extracted from automated deep learning (DL) segmentations against those derived from physician annotations. Model performance was assessed using the concordance index (C-index). Results: Among all types of features, the CoxNet model using conventional baseline PET metrics achieved the highest performance (C-index, 0.72 ± 0.01), significantly outperforming the clinical model (C-index, 0.65 ± 0.01). Neither the inclusion of radiomics features nor interim PET-derived features improved performance. Models using DL-generated segmentations achieved comparable prognostic accuracy (C-index, 0.72 ± 0.01) to those using physician-based segmentations. Conclusion: Quantitative PET features provided significant prognostic improvements over clinical variables. Furthermore, DL-based segmentation offered a promising approach for automating feature extraction, supporting more effective risk stratification in pediatric HL.

Vision-language models can connect the text description of an object to its specific location in an image through visual grounding. This has potential applications in enhanced radiology reporting. However, these models require large annotated image-text datasets, which are lacking for PET/CT. We developed an automated pipeline to generate weak image-text labels and used it to train a 3D visual grounding model. Our weak-labeling pipeline identified sentences describing positive findings in PET/CT reports by searching for mentions of standardized uptake values (SUVmax) and axial slice numbers. These were used to automatically generate lesion masks, which were paired with the corresponding text descriptions. From 25,578 PET/CT exams, we extracted 11,356 sentence-label pairs. Using this data, we trained ConTEXTual Net 3D, which takes as input a description of a lesion and generates a corresponding segmentation mask. The model’s performance was evaluated on 251 radiologist-reviewed cases and compared against LLMSeg, a 2.5D version of ConTEXTual Net, and two radiologists. We evaluated detection performance using F1 score. The weak-labeling pipeline accurately identified lesion locations in 98% of cases (246/251). ConTEXTual Net 3D achieved an F1 score of 0.80, outperforming LLMSeg (F1 = 0.22) and the 2.5D model (F1 = 0.53), though it underperformed both radiologists (F1 = 0.94 and 0.91). The model achieved better performance on ¹⁸ F-fluorodeoxyglucose (F1 = 0.78) and DCFPyL (F1 = 0.75) exams than on DOTATATE (F1 = 0.58) and ¹⁸ F-fluciclovine (F1 = 0.66) exams. In conclusion, our novel weak labeling pipeline accurately produced an annotated dataset of PET/CT image-text pairs. ConTEXTual Net 3D significantly outperformed other models but fell short of the performance of nuclear medicine physicians. Our study suggests that even larger datasets may be needed to close this performance gap.

There is a pressing need for improved standardization of terminology and data in nuclear medicine. The field is experiencing unprecedented growth, driven by advances in radiopharmaceutical therapy (RPT) and the emergence of artificial intelligence (AI). However, there are challenges that threaten to frustrate this continued progress. For instance, despite the successes of RPT, high-quality evidence on how to best personalize RPT and take full advantage of its theranostics properties is still lacking. To obtain this evidence, large, structured datasets are needed to associate different RPT strategies with patient outcomes. Large datasets are also needed for the development of AI algorithms, especially as new foundation models demand increasingly large training datasets. Both of these obstacles could be overcome by multiinstitutional data sharing. However, inconsistencies in terminology and data collection make effective data pooling difficult. This article, produced by the Society of Nuclear Medicine and Molecular Imaging AI–Dosimetry Working Group, discusses the need for standardization in nuclear medicine terminology and data. We advocate for the adoption of standardized data and metadata frameworks based on controlled biomedical ontologies to better harmonize the collection of nuclear medicine data. We provide recommendations for the field that, if followed, would facilitate multiinstitutional data sharing and allow for the collection of large datasets. We describe a use case demonstrating how standardized vocabularies and data collection can enhance efforts to associate theranostics target expression data with patient outcomes.

A longitudinally aware segmentation network trained on multicenter clinical trial data achieved high performance in automatic quantification of PET metrics for baseline and interim-therapy scans in pediatric Hodgkin lymphoma.

Accurate segmentation of gross tumor volume (GTV) is essential for effective MRI-guided adaptive radiotherapy (MRgART) in head and neck cancer. However, manual segmentation of the GTV over the course of therapy is time-consuming and prone to interobserver variability. Deep learning (DL) has the potential to overcome these challenges by automatically delineating GTVs. In this study, our team, UW LAIR, tackled the challenges of both pre-radiotherapy (pre-RT) (Task 1) and mid-radiotherapy (mid-RT) (Task 2) tumor volume segmentation. To this end, we developed a series of DL models for longitudinal GTV segmentation. The backbone of our models for both tasks was SegResNet with deep supervision. For Task 1, we trained the model using a combined dataset of pre-RT and mid-RT MRI data, which resulted in the improved aggregated Dice similarity coefficient (DSC_agg) on a hold-out internal testing set compared to models trained solely on pre-RT MRI data. In Task 2, we introduced mask-aware attention modules, enabling pre-RT GTV masks to influence intermediate features learned from mid-RT data. This attention-based approach yielded slight improvements over the baseline method, which concatenated mid-RT MRI with pre-RT GTV masks as input. In the final testing phase, the ensemble of 10 pre-RT segmentation models achieved an average DSC_agg of 0.794, with 0.745 for primary GTV (GTVp) and 0.844 for metastatic lymph nodes (GTVn) in Task 1. For Task 2, the ensemble of 10 mid-RT segmentation models attained an average DSC_agg of 0.733, with 0.607 for GTVp and 0.859 for GTVn, leading us to achieve 1st place. In summary, we presented a collection of DL models that could facilitate GTV segmentation in MRgART, offering the potential to streamline radiation oncology workflows.

Large language models (LLMs) are poised to have a disruptive impact on health care. Numerous studies have demonstrated promising applications of LLMs in medical imaging, and this number will grow as LLMs further evolve into large multimodal models (LMMs) capable of processing both text and images. Given the substantial roles that LLMs and LMMs will have in health care, it is important for physicians to understand the underlying principles of these technologies so they can use them more effectively and responsibly and help guide their development. This article explains the key concepts behind the development and application of LLMs, including token embeddings, transformer networks, self-supervised pretraining, fine-tuning, and others. It also describes the technical process of creating LMMs and discusses use cases for both LLMs and LMMs in medical imaging.

Objective. In this work, we present and evaluate a technique for performing interface measurements of beta particle-emitting radiopharmaceutical therapy agents in solution. Approach. Unlaminated EBT3 film was calibrated for absorbed dose to water using a NIST matched x-ray beam. Custom acrylic source phantoms were constructed and placed above interfaces comprised of bone, lung, and water-equivalent materials. The film was placed perpendicular to these interfaces and measurements for absorbed dose to water using solutions of ⁹⁰Y and ¹⁷⁷Lu were performed and compared to Monte Carlo absorbed dose to water estimates simulated with EGSnrc. Surface and depth dose profile measurements were also performed. Main results. Surface absorbed dose to water measurements agreed with predicted results within 3.6% for ¹⁷⁷Lu and 2.2% for ⁹⁰Y. The agreement between predicted and measured absorbed dose to water was better for ⁹⁰Y than ¹⁷⁷Lu for depth dose and interface profiles. In general, agreement within k = 1 uncertainty bounds was observed for both radionuclides and all interfaces. An exception to this was found for the bone-to-water interface for ¹⁷⁷Lu due to the increased sensitivity of the measurements to imperfections in the material surfaces. Significance. This work demonstrates the feasibility and limitations of using radiochromic film for performing absorbed dose to water measurements on beta particle-emitting radiopharmaceutical therapy agents across material interfaces.

Objective . This work introduces a novel approach to performing active and passive dosimetry for beta-emitting radionuclides in solution using common dosimeters. The measurements are compared to absorbed dose to water ( D _w ) estimates from Monte Carlo (MC) simulations. We present a method for obtaining absorbed dose to water, measured with dosimeters, from beta-emitting radiopharmaceutical agents using a custom SPECT/CT compatible phantom for validation of Monte Carlo based absorbed dose to water estimates. Approach . A cylindrical, acrylic SPECT/CT compatible phantom capable of housing an IBA EFD diode, Exradin A20-375 parallel plate ion chamber, unlaminated EBT3 film, and thin TLD100 microcubes was constructed for the purpose of measuring absorbed dose to water from solutions of common beta-emitting radiopharmaceutical therapy agents. The phantom is equipped with removable detector inserts that allow for multiple configurations and is designed to be used for validation of image-based absorbed dose estimates with detector measurements. Two experiments with ¹³¹ I and one experiment with ¹⁷⁷ Lu were conducted over extended measurement intervals with starting activities of approximately 150–350 MBq. Measurement data was compared to Monte Carlo simulations using the egs_chamber user code in EGSnrc 2019. Main results . Agreement within k = 1 uncertainty between measured and MC predicted D _w was observed for all dosimeters, except the A20-375 ion chamber during the second ¹³¹ I experiment. Despite the agreement, the measured values were generally lower than predicted values by 5%–15%. The uncertainties at k = 1 remain large (5%–30% depending on the dosimeter) relative to other forms of radiation therapy. Significance . Despite high uncertainties, the overall agreement between measured and simulated absorbed doses is promising for the use of dosimeter-based RPT measurements in the validation of MC predicted D _w .