Search

Abstract Healthcare datasets are becoming larger and more complex, necessitating the development of accurate and generalizable AI models for medical applications. Unstructured datasets, including medical imaging, electrocardiograms, and natural language data, are gaining attention with advancements in deep convolutional neural networks and large language models. However, estimating the generalizability of these models to new healthcare settings without extensive validation on external data remains challenging. In experiments across 13 datasets including X-rays, CTs, ECGs, clinical discharge summaries, and lung auscultation data, our results demonstrate that model performance is frequently overestimated by up to 20% on average due to shortcut learning of hidden data acquisition biases (DAB). Shortcut learning refers to a phenomenon in which an AI model learns to solve a task based on spurious correlations present in the data as opposed to features directly related to the task itself. We propose an open source, bias-corrected external accuracy estimate, P E s t , that better estimates external accuracy to within 4% on average by measuring and calibrating for DAB-induced shortcut learning.

Background and purpose: No best practices currently exist for achieving high quality radiation therapy (RT) treatment plan adaptation during magnetic resonance (MR) guided RT of prostate cancer. This study validates the use of machine learning (ML) automated RT treatment plan adaptation and benchmarks it against current clinical RT plan adaptation methods. Materials and methods: We trained an atlas-based ML automated treatment planning model using reference MR RT treatment plans (42.7 Gy in 7 fractions) from 46 patients with prostate cancer previously treated at our institution. For a held-out test set of 38 patients, retrospectively generated ML RT plans were compared to clinical human-generated adaptive RT plans for all 266 fractions. Differences in dose-volume metrics and clinical objective pass rates were evaluated using Wilcoxon tests (p

Abstract Purpose High-quality radiotherapy (RT) planning for children and young adults with primary brain tumours is essential to minimize the risk of late treatment effects. The feasibility of using automated machine-learning (ML) to aid RT planning in this population has not previously been studied. Methods and materials We developed a ML model that identifies learned relationships between image features and expected dose in a training set of 95 patients with a primary brain tumour treated with focal radiotherapy to a dose of 54 Gy in 30 fractions. This ML method was then used to create predicted dose distributions for 15 previously-treated brain tumour patients across two institutions, as a testing set. Dosimetry to target volumes and organs-at-risk (OARs) were compared between the clinically-delivered (human-generated) plans versus the ML plans. Results The ML method was able to create deliverable plans in all 15 patients in the testing set. All ML plans were generated within 30 min of initiating planning. Planning target volume coverage with 95% of the prescription dose was attained in all plans. OAR doses were similar across most structures evaluated; mean doses to brain and left temporal lobe were lower in ML plans than manual plans (mean difference to left temporal, – 2.3 Gy, p = 0.006; mean differences to brain, – 1.3 Gy, p = 0.017), whereas mean doses to right cochlea and lenses were higher in ML plans (+ 1.6–2.2 Gy, p

Summary: Background: Endothelial cell (EC) activation, endotheliitis, vascular permeability, and thrombosis have been observed in patients with severe coronavirus disease 2019 (COVID-19), indicating that the vasculature is affected during the acute stages of SARS-CoV-2 infection. It remains unknown whether circulating vascular markers are sufficient to predict clinical outcomes, are unique to COVID-19, and if vascular permeability can be therapeutically targeted. Methods: Prospectively evaluating the prevalence of circulating inflammatory, cardiac, and EC activation markers as well as developing a microRNA atlas in 241 unvaccinated patients with suspected SARS-CoV-2 infection allowed for prognostic value assessment using a Random Forest model machine learning approach. Subsequent ex vivo experiments assessed EC permeability responses to patient plasma and were used to uncover modulated gene regulatory networks from which rational therapeutic design was inferred. Findings: Multiple inflammatory and EC activation biomarkers were associated with mortality in COVID-19 patients and in severity-matched SARS-CoV-2-negative patients, while dysregulation of specific microRNAs at presentation was specific for poor COVID-19-related outcomes and revealed disease-relevant pathways. Integrating the datasets using a machine learning approach further enhanced clinical risk prediction for in-hospital mortality. Exposure of ECs to COVID-19 patient plasma resulted in severity-specific gene expression responses and EC barrier dysfunction, which was ameliorated using angiopoietin-1 mimetic or recombinant Slit2-N. Interpretation: Integration of multi-omics data identified microRNA and vascular biomarkers prognostic of in-hospital mortality in COVID-19 patients and revealed that vascular stabilizing therapies should be explored as a treatment for endothelial dysfunction in COVID-19, and other severe diseases where endothelial dysfunction has a central role in pathogenesis. Funding Information: This work was directly supported by grant funding from the Ted Rogers Center for Heart Research, Toronto, Ontario, Canada and the Peter Munk Cardiac Center, Toronto, Ontario, Canada.

Quality assurance of data prior to use in automated pipelines and image analysis would assist in safeguarding against biases and incorrect interpretation of results. Automation of quality assurance steps would further improve robustness and efficiency of these methods, motivating widespread adoption of techniques. Previous work by our group demonstrated the ability of convolutional neural networks (CNN) to efficiently classify head and neck (H&N) computed-tomography (CT) images for the presence of dental artifacts (DA) that obscure visualization of structures and the accuracy of Hounsfield units. In this work we demonstrate the generalizability of our previous methodology by validating CNNs on six external datasets, and the potential benefits of transfer learning with fine-tuning on CNN performance. 2112 H&N CT images from seven institutions were scored as DA positive or negative. 1538 images from a single institution were used to train three CNNs with resampling grid sizes of 643, 1283 and 2563. The remaining six external datasets were used in five-fold cross-validation with a data split of 20% training/fine-tuning and 80% validation. The three pre-trained models were each validated using the five-folds of the six external datasets. The pre-trained models also underwent transfer learning with fine-tuning using the 20% training/fine-tuning data, and validated using the corresponding validation datasets. The highest micro-averaged AUC for our pre-trained models across all external datasets occurred with a resampling grid of 2563 (AUC  =  0.91  ±  0.01). Transfer learning with fine-tuning improved generalizability when utilizing a resampling grid of 2563 to a micro-averaged AUC of 0.92  ±  0.01. Despite these promising results, transfer learning did not improve AUC when utilizing small resampling grids or small datasets. Our work demonstrates the potential of our previously developed automated quality assurance methods to generalize to external datasets. Additionally, we showed that transfer learning with fine-tuning using small portions of external datasets can be used to fine-tune models for improved performance when large variations in images are present. [ABSTRACT FROM AUTHOR]

Recent works in automated radiotherapy treatment planning have used machine learning based on historical treatment plans to infer the spatial dose distribution for a novel patient directly from the planning image. We present a probabilistic, atlas-based approach which predicts the dose for novel patients using a set of automatically selected most similar patients (atlases). The output is a spatial dose objective, which specifies the desired dose-per-voxel, and therefore replaces the need to specify and tune dose-volume objectives. Voxel-based dose mimicking optimization then converts the predicted dose distribution to a complete treatment plan with dose calculation using a collapsed cone convolution dose engine. In this study, we investigated automated planning for right-sided oropharaynx head and neck patients treated with IMRT and VMAT. We compare four versions of our dose prediction pipeline using a database of 54 training and 12 independent testing patients by evaluating 14 clinical dose evaluation criteria. Our preliminary results are promising and demonstrate that automated methods can generate comparable dose distributions to clinical. Overall, automated plans achieved an average of 0.6% higher dose for target coverage evaluation criteria, and 2.4% lower dose at the organs at risk criteria levels evaluated compared with clinical. There was no statistically significant difference detected in high-dose conformity between automated and clinical plans as measured by the conformation number. Automated plans achieved nine more unique criteria than clinical across the 12 patients tested and automated plans scored a significantly higher dose at the evaluation limit for two high-risk target coverage criteria and a significantly lower dose in one critical organ maximum dose. The novel dose prediction method with dose mimicking can generate complete treatment plans in 12–13 min without user interaction. It is a promising approach for fully automated treatment planning and can be readily applied to different treatment sites and modalities. [ABSTRACT FROM AUTHOR]

Automating the radiotherapy treatment planning process is a technically challenging problem. The majority of automated approaches have focused on customizing and inferring dose volume objectives to be used in plan optimization. In this work we outline a multi-patient atlas-based dose prediction approach that learns to predict the dose-per-voxel for a novel patient directly from the computed tomography planning scan without the requirement of specifying any objectives. Our method learns to automatically select the most effective atlases for a novel patient, and then map the dose from those atlases onto the novel patient. We extend our previous work to include a conditional random field for the optimization of a joint distribution prior that matches the complementary goals of an accurately spatially distributed dose distribution while still adhering to the desired dose volume histograms. The resulting distribution can then be used for inverse-planning with a new spatial dose objective, or to create typical dose volume objectives for the canonical optimization pipeline. We investigated six treatment sites (633 patients for training and 113 patients for testing) and evaluated the mean absolute difference in all DVHs for the clinical and predicted dose distribution. The results on average are favorable in comparison to our previous approach (1.91 versus 2.57). Comparing our method with and without atlas-selection further validates that atlas-selection improved dose prediction on average in whole breast (0.64 versus 1.59), prostate (2.13 versus 4.07), and rectum (1.46 versus 3.29) while it is less important in breast cavity (0.79 versus 0.92) and lung (1.33 versus 1.27) for which there is high conformity and minimal dose shaping. In CNS brain, atlas-selection has the potential to be impactful (3.65 versus 5.09), but selecting the ideal atlas is the most challenging. [ABSTRACT FROM AUTHOR]