Uncovering spatiotemporal dynamics of the corticothalamic network at ictal onset.

OBJECTIVE: Although the clinical efficacy of deep brain stimulation targeting the anterior nucleus (AN) and centromedian nucleus (CM) of the thalamus has been actively investigated for the treatment of medication-resistant epilepsy, few studies have investigated dynamic ictal changes in corticothalamic connectivity in human electroencephalographic (EEG) recording. This study aims to establish the complex spatiotemporal dynamics of the ictal corticothalamic network associated with various seizure foci. METHODS: We analyzed 10 patients (aged 2.7-28.1 years) with medication-resistant focal epilepsy who underwent stereotactic EEG evaluation with thalamic sampling. We examined both undirected and directed connectivity, incorporating coherence and spectral Granger causality analysis (GCA) between the diverse seizure foci and thalamic nuclei (AN and CM) at ictal onset. RESULTS: In our analysis of 36 seizures, coherence between seizure onset and thalamic nuclei increased across all frequencies, especially in slower bands (delta, theta, alpha). GCA showed increased information flow from seizure onset to the thalamus across all frequency bands, but outflows from the thalamus were mainly in slower frequencies, particularly delta. In the subgroup analysis based on various seizure foci, the delta coherence showed a more pronounced increase at CM than at AN during frontal lobe seizures. Conversely, in limbic seizures, the delta coherence increase was greater at AN compared to CM. SIGNIFICANCE: It appears that the delta frequency plays a pivotal role in modulating the corticothalamic network during seizures. Our results underscore the significance of comprehending the spatiotemporal dynamics of the corticothalamic network at ictal onset, and this knowledge could guide personalized responsive neuromodulation treatment strategies.

Automated prostate gland segmentation in challenging clinical cases: comparison of three artificial intelligence methods.

OBJECTIVE: Automated methods for prostate segmentation on MRI are typically developed under ideal scanning and anatomical conditions. This study evaluates three different prostate segmentation AI algorithms in a challenging population of patients with prior treatments, variable anatomic characteristics, complex clinical history, or atypical MRI acquisition parameters. MATERIALS AND METHODS: A single institution retrospective database was queried for the following conditions at prostate MRI: prior prostate-specific oncologic treatment, transurethral resection of the prostate (TURP), abdominal perineal resection (APR), hip prosthesis (HP), diversity of prostate volumes (large ≥ 150 cc, small ≤ 25 cc), whole gland tumor burden, magnet strength, noted poor quality, and various scanners (outside/vendors). Final inclusion criteria required availability of axial T2-weighted (T2W) sequence and corresponding prostate organ segmentation from an expert radiologist. Three previously developed algorithms were evaluated: (1) deep learning (DL)-based model, (2) commercially available shape-based model, and (3) federated DL-based model. Dice Similarity Coefficient (DSC) was calculated compared to expert. DSC by model and scan factors were evaluated with Wilcox signed-rank test and linear mixed effects (LMER) model. RESULTS: 683 scans (651 patients) met inclusion criteria (mean prostate volume 60.1 cc [9.05-329 cc]). Overall DSC scores for models 1, 2, and 3 were 0.916 (0.707-0.971), 0.873 (0-0.997), and 0.894 (0.025-0.961), respectively, with DL-based models demonstrating significantly higher performance (p < 0.01). In sub-group analysis by factors, Model 1 outperformed Model 2 (all p < 0.05) and Model 3 (all p < 0.001). Performance of all models was negatively impacted by prostate volume and poor signal quality (p < 0.01). Shape-based factors influenced DL models (p < 0.001) while signal factors influenced all (p < 0.001). CONCLUSION: Factors affecting anatomical and signal conditions of the prostate gland can adversely impact both DL and non-deep learning-based segmentation models.

The effect of dose-interval on antibody response to mRNA COVID-19 vaccines: a prospective cohort study.

Background: Vaccination against COVID-19 is highly effective in preventing severe disease and hospitalization, but primary COVID mRNA vaccination schedules often differed from those recommended by the manufacturers due to supply chain issues. We investigated the impact of delaying the second dose on antibody responses to COVID mRNA-vaccines in a prospective cohort of health-care workers in Quebec. Methods: We recruited participants from the McGill University Health Centre who provided serum or participant-collected dried blood samples (DBS) at 28-days, 3 months, and 6 months post-second dose and at 28-days after a third dose. IgG antibodies to SARS-CoV2 spike (S), the receptor-binding domain (RBD), nucleocapsid (N) and neutralizing antibodies to the ancestral strain were assessed by enzyme-linked immunosorbent assay (ELISA). We examined associations between long (≤89 days) versus short (<89 days) between-dose intervals and antibody response through multivariable mixed-effects models adjusted for age, sex, prior covid infection status, time since vaccine dose, and assay batch. Findings: The cohort included 328 participants who received up to three vaccine doses (>80% Pfizer-BioNTech). Weighted averages of the serum (n=744) and DBS (n=216) cohort results from the multivariable models showed that IgG anti-S was 31% higher (95% CI: 12% to 53%) and IgG anti-RBD was 37% higher (95% CI: 14% to 65%) in the long vs. short interval participants, across all time points. Interpretation: Our study indicates that extending the covid primary series between-dose interval beyond 89 days (approximately 3 months) provides stronger antibody responses than intervals less than 89 days. Our demonstration of a more robust antibody response with a longer between dose interval is reassuring as logistical and supply challenges are navigated in low-resource settings.

A multitask approach for automated detection and segmentation of thyroid nodules in ultrasound images.

An increase in the incidence and diagnosis of thyroid nodules and thyroid cancer underscores the need for a better approach to nodule detection and risk stratification in ultrasound (US) images that can reduce healthcare costs, patient discomfort, and unnecessary invasive procedures. However, variability in ultrasound technique and interpretation makes the diagnostic process partially subjective. Therefore, an automated approach that detects and segments nodules could improve performance on downstream tasks, such as risk stratification. Ultrasound studies were acquired from 280 patients at UCLA Health, totaling 9888 images, and annotated by collaborating radiologists. Current deep learning architectures for segmentation are typically semi-automated because they are evaluated solely on images known to have nodules and do not assess ability to identify suspicious images. However, the proposed multitask approach both detects suspicious images and segments potential nodules; this allows for a clinically translatable model that aptly parallels the workflow for thyroid nodule assessment. The multitask approach is centered on an anomaly detection (AD) module that can be integrated with any UNet architecture variant to improve image-level nodule detection. Of the evaluated multitask models, a UNet with a ImageNet pretrained encoder and AD achieved the highest F1 score of 0.839 and image-wide Dice similarity coefficient of 0.808 on the hold-out test set. Furthermore, models were evaluated on two external validations datasets to demonstrate generalizability and robustness to data variability. Ultimately, the proposed architecture is an automated multitask method that expands on previous methods by successfully both detecting and segmenting nodules in ultrasound.

Perspectives, experiences, and concerns with cyclical vomiting syndrome: Insights from online targeted-disease forums.

INTRODUCTION: There is limited research examining the biopsychosocial impact of cyclic vomiting syndrome (CVS) on patients. This study aims to assess individuals' experiences, fears, and concerns associated with CVS and the impact of CVS on their daily lives. METHODS: We employed social netnography to analyze publicly available posts related to CVS that were identified from six US online forums and Twitter. A randomly selected sub-cohort of posts per pre-defined criteria was first qualitatively analyzed using an inductive thematic approach. Then, machine learning topic modeling was applied to explore themes in an unsupervised manner for the entire corpus of posts. Afterward, findings from the qualitative and quantitative approaches were integrated to generate a thematic network. RESULTS: Based on the 39,179 collected posts, seven domain themes were identified. Overall, 41.4% of the posts were related to "biopsychosocial burden" of CVS, including physical impact, psychological impact, and social impact. In 22.3% of posts, individuals shared their experience of "interactions with the healthcare system", and 14.2% of posts were related to "perceived CVS triggers." Individuals also shared "solutions to alleviate their symptoms" and "mental health needs" in 10.2% and 8.8% of posts, respectively. Finally, 6.1% of the posts were about "seeking/sharing support" with others. DISCUSSION: This is the first social netnography study to describe the in-depth experiences of individuals living with CVS and the marked impact on their physical, mental, and social health. The study also highlights the unmet need for effective therapies, both pharmacological and non-pharmacological, to alleviate the biopsychosocial impact of CVS.

SLIViT: a general AI framework for clinical-feature diagnosis from limited 3D biomedical-imaging data.

We present SLIViT, a deep-learning framework that accurately measures disease-related risk factors in volumetric biomedical imaging, such as magnetic resonance imaging (MRI) scans, optical coherence tomography (OCT) scans, and ultrasound videos. To evaluate SLIViT, we applied it to five different datasets of these three different data modalities tackling seven learning tasks (including both classification and regression) and found that it consistently and significantly outperforms domain-specific state-of-the-art models, typically improving performance (ROC AUC or correlation) by 0.1-0.4. Notably, compared to existing approaches, SLIViT can be applied even when only a small number of annotated training samples is available, which is often a constraint in medical applications. When trained on less than 700 annotated volumes, SLIViT obtained accuracy comparable to trained clinical specialists while reducing annotation time by a factor of 5,000 demonstrating its utility to automate and expedite ongoing research and other practical clinical scenarios.

Patients' Perspectives, Experiences, and Concerns With Perianal Fistulae: Insights From Online Targeted-Disease Forums.

Background: Perianal fistulae can undermine physical, emotional, and social well-being in patients with Crohn's disease and are challenging to manage. Social media offers a rich opportunity to gain an in-depth understanding of the impact of perianal fistulae on patients' daily lives outside of controlled environments. In this study, we conducted social media analytics to examine patients' experiences with perianal fistulae and assessed the impact of perianal fistulae on patients' behavior and overall well-being. Methods: We used a mixed-method approach to examine 119 986 publicly available posts collected from 10 Crohn's disease forums in the United States between January 01, 2010 and January 01, 2020. Discussions related to Crohn's perianal fistulae were retrieved. We randomly selected 700 posts and qualitatively analyzed them using an inductive thematic approach. We then applied a latent Dirichlet allocation probabilistic topic model to explore themes in an unsupervised manner on the collection of 119 986 posts. Results: In the qualitative analysis, 5 major themes were identified: (1) burden of perianal fistula; (2) challenges associated with treatment; (3) online information seeking and sharing; (4) patient experiences with treatments; and (5) patients' apprehension about treatments. In the quantitative analysis, the percentages of posts related to the major themes were (1) 20%, (2) 29%, (3) 66%, and (4) 28%, while the topic model did not identify theme 5. Conclusions: Social media reveals a dynamic range of themes governing patients' perspectives and experiences with Crohn's perianal fistulae. In addition to the biopsychosocial burden, patients frequently express dissatisfaction with current treatments and often struggle to navigate among available management options.

Latent diffusion augmentation enhances deep learning analysis of neuro-morphology in limbal stem cell deficiency.

Introduction: Limbal Stem Cell Deficiency (LSCD) is a blinding corneal disease characterized by the loss of function or deficiency in adult stem cells located at the junction between the cornea and the sclera (i.e., the limbus), namely the limbal stem cells (LSCs). Recent advances in in vivo imaging technology have improved disease diagnosis and staging to quantify several biomarkers of in vivo LSC function including epithelial thickness measured by anterior segment optical coherence tomography, and basal epithelial cell density and subbasal nerve plexus by in vivo confocal microscopy. A decrease in central corneal sub-basal nerve density and nerve fiber and branching number has been shown to correlate with the severity of the disease in parallel with increased nerve tortuosity. Yet, image acquisition and manual quantification require a high level of expertise and are time-consuming. Manual quantification presents inevitable interobserver variability. Methods: The current study employs a novel deep learning approach to classify neuron morphology in various LSCD stages and healthy controls, by integrating images created through latent diffusion augmentation. The proposed model, a residual U-Net, is based in part on the InceptionResNetV2 transfer learning model. Results: Deep learning was able to determine fiber number, branching, and fiber length with high accuracy (R2 of 0.63, 0.63, and 0.80, respectively). The model trained on images generated through latent diffusion on average outperformed the same model when trained on solely original images. The model was also able to detect LSCD with an AUC of 0.867, which showed slightly higher performance compared to classification using manually assessed metrics. Discussion: The results suggest that utilizing latent diffusion to supplement training data may be effective in bolstering model performance. The results of the model emphasize the ability as well as the shortcomings of this novel deep learning approach to predict various nerve morphology metrics as well as LSCD disease severity.

Results of the Stop the Spread Ottawa (SSO) cohort study: a Canadian urban-based prospective evaluation of antibody responses and neutralisation efficiency to SARS-CoV-2 infection and vaccination.

BACKGROUND: Predictors of COVID-19 vaccine immunogenicity and the influence of prior severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection require elucidation. METHODS: Stop the Spread Ottawa is a prospective cohort of individuals at-risk for or who have been infected with SARS-CoV-2, initially enrolled for 10 months beginning October 2020. This cohort was enriched for public-facing workers. This analysis focuses on safety and immunogenicity of the initial two doses of COVID-19 vaccine. RESULTS: Post-vaccination data with blood specimens were available for 930 participants. 22.8% were SARS-CoV2 infected prior to the first vaccine dose. Cohort characteristics include: median age 44 (IQR: 22-56), 66.6% women, 89.0% white, 83.2% employed. 38.1% reported two or more comorbidities and 30.8% reported immune compromising condition(s). Over 95% had detectable IgG levels against the spike and receptor binding domain (RBD) 3 months post second vaccine dose. By multivariable analysis, increasing age and high-level immune compromise predicted diminishing IgG spike and RBD titres at month 3 post second dose. IgG spike and RBD titres were higher immediately post vaccination in those with SARS-CoV-2 infection prior to first vaccination and spike titres were higher at 6 months in those with wider time intervals between dose 1 and 2. IgG spike and RBD titres and neutralisation were generally similar by sex, weight and whether receiving homogeneous or heterogeneous combinations of vaccines. Common symptoms post dose 1 vaccine included fatigue (64.7%), injection site pain (47.5%), headache (27.2%), fever/chills (26.2%) and body aches (25.3%). These symptoms were similar with subsequent doses. CONCLUSION: The initial two COVID-19 vaccine doses are safe, well-tolerated and highly immunogenic across a broad spectrum of vaccine recipients including those working in public facing environments.

Cohort profile: investigating SARS-CoV-2 infection and the health and psychosocial impact of the COVID-19 pandemic in the Canadian CHILD Cohort.

The coronavirus disease 2019 (COVID-19) pandemic has affected all Canadian families, with some impacted differently than others. Our study aims to: (1) determine the prevalence and transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection among Canadian families, (2) identify predictors of infection susceptibility and severity of SARS-CoV-2, and (3) identify health and psychosocial impacts of the COVID-19 pandemic. This study builds upon the CHILD Cohort Study, an ongoing multi-ethnic general population prospective cohort consisting of 3,454 Canadian families with children born in Vancouver, Edmonton, Manitoba, and Toronto between 2009 and 2012. During the pandemic, CHILD households were invited to participate in the CHILD COVID-19 Add-On Study involving: (1) brief biweekly surveys about COVID-19 symptoms and testing; (2) quarterly questionnaires assessing COVID-19 exposure and testing, vaccination status, physical and mental health, and pandemic-driven life changes; and (3) in-home biological sampling kits to collect blood and stool. In total, 1,462 households (5,378 participants) consented to the CHILD COVID-19 Add-On Study: 2,803 children (mean±standard deviation [SD], 9.0±2.7 years; range, 0-17 years) and 2,576 adults (mean±SD, 43.0±6.5 years; range, 18-85 years). We will leverage the wealth of pre-pandemic CHILD data to identify risk and resilience factors for susceptibility and severity to the direct and indirect pandemic effects. Our short-term findings will inform key stakeholders and knowledge users to shape current and future pandemic responses. Additionally, this study provides a unique resource to study the long-term impacts of the pandemic as the CHILD Cohort Study continues.

Perfusion Collateral Index versus Hypoperfusion Intensity Ratio in Assessment of Collaterals in Patients with Acute Ischemic Stroke.

BACKGROUND AND PURPOSE: Perfusion-based collateral indices such as the perfusion collateral index and the hypoperfusion intensity ratio have shown promise in the assessment of collaterals in patients with acute ischemic stroke. We aimed to compare the diagnostic performance of the perfusion collateral index and the hypoperfusion intensity ratio in collateral assessment compared with angiographic collaterals and outcome measures, including final infarct volume, infarct growth, and functional independence. MATERIALS AND METHODS: Consecutive patients with acute ischemic stroke with anterior circulation proximal arterial occlusion who underwent endovascular thrombectomy and had pre- and posttreatment MRI were included. Using pretreatment MR perfusion, we calculated the perfusion collateral index and the hypoperfusion intensity ratio for each patient. The angiographic collaterals obtained from DSA were dichotomized to sufficient (American Society of Interventional and Therapeutic Neuroradiology [ASITN] scale 3-4) versus insufficient (ASITN scale 0-2). The association of collateral status determined by the perfusion collateral index and the hypoperfusion intensity ratio was assessed against angiographic collaterals and outcome measures. RESULTS: A total of 98 patients met the inclusion criteria. Perfusion collateral index values were significantly higher in patients with sufficient angiographic collaterals (P < .001), while there was no significant (P = .46) difference in hypoperfusion intensity ratio values. Among patients with good (mRS 0-2) versus poor (mRS 3-6) functional outcome, the perfusion collateral index of ≥ 62 was present in 72% versus 31% (P = .003), while the hypoperfusion intensity ratio of ≤0.4 was present in 69% versus 56% (P = .52). The perfusion collateral index and the hypoperfusion intensity ratio were both significantly predictive of final infarct volume, but only the perfusion collateral index was significantly (P = .03) associated with infarct growth. CONCLUSIONS: Results show that the perfusion collateral index outperforms the hypoperfusion intensity ratio in the assessment of collateral status, infarct growth, and determination of functional outcomes.

Uncovering Spatiotemporal Dynamics of the Corticothalamic Network during Seizures.

Objective: Although the clinical efficacy of deep brain stimulation targeting the anterior nucleus (AN) and centromedian nucleus (CM) of the thalamus has been actively investigated for the treatment of medication-resistant epilepsy, few studies have investigated dynamic ictal changes in corticothalamic connectivity in human EEG recording. This study aims to establish the complex spatiotemporal dynamics of the ictal corticothalamic network associated with various seizure foci. Methods: We analyzed ten patients (aged 2.7-28.1) with medication-resistant focal epilepsy who underwent stereotactic EEG evaluation with thalamic coverage. We examined both undirected and directed connectivity, incorporating coherence and spectral Granger causality analysis (GCA) between the diverse seizure foci and thalamic nuclei (AN and CM). Results: In our analysis of 36 seizures, coherence between seizure onset and thalamic nuclei increased across all frequencies, especially in slower bands (delta, theta, alpha). GCA showed increased information flow from seizure onset to the thalamus across all frequency bands, but outflows from the thalamus were mainly in slower frequencies, particularly delta. In the subgroup analysis based on various seizure foci, the delta coherence showed a more pronounced increase at CM than at AN during frontal lobe seizures. Conversely, in limbic seizures, the delta coherence increase was greater at AN compared to CM. Interpretation: It appears that the delta frequency plays a pivotal role in modulating the corticothalamic network during seizures. Our results underscore the significance of comprehending the spatiotemporal dynamics of the corticothalamic network during seizures, and this knowledge could guide personalized neuromodulation treatment strategies.

Multi-site, Multi-domain Airway Tree Modeling.

Open international challenges are becoming the de facto standard for assessing computer vision and image analysis algorithms. In recent years, new methods have extended the reach of pulmonary airway segmentation that is closer to the limit of image resolution. Since EXACT'09 pulmonary airway segmentation, limited effort has been directed to the quantitative comparison of newly emerged algorithms driven by the maturity of deep learning based approaches and extensive clinical efforts for resolving finer details of distal airways for early intervention of pulmonary diseases. Thus far, public annotated datasets are extremely limited, hindering the development of data-driven methods and detailed performance evaluation of new algorithms. To provide a benchmark for the medical imaging community, we organized the Multi-site, Multi-domain Airway Tree Modeling (ATM'22), which was held as an official challenge event during the MICCAI 2022 conference. ATM'22 provides large-scale CT scans with detailed pulmonary airway annotation, including 500 CT scans (300 for training, 50 for validation, and 150 for testing). The dataset was collected from different sites and it further included a portion of noisy COVID-19 CTs with ground-glass opacity and consolidation. Twenty-three teams participated in the entire phase of the challenge and the algorithms for the top ten teams are reviewed in this paper. Both quantitative and qualitative results revealed that deep learning models embedded with the topological continuity enhancement achieved superior performance in general. ATM'22 challenge holds as an open-call design, the training data and the gold standard evaluation are available upon successful registration via its homepage (https://atm22.grand-challenge.org/).

Antibody neutralization capacity after coronavirus disease 2019 vaccination in people with HIV in Canada.

OBJECTIVES: Many vaccines require higher/additional doses or adjuvants to provide adequate protection for people with HIV (PWH). Here, we compare coronavirus disease 2019 (COVID-19) vaccine-induced antibody neutralization capacity in PWH vs. HIV-negative individuals following two vaccine doses. DESIGN: In Canadian prospective observational cohorts, including a multicentre study of PWH receiving at least two COVID-19 vaccinations (mRNA or ChAdOx1-S), and a parallel study of HIV-negative controls (Stop the Spread Ottawa Cohort), we measured vaccine-induced neutralization capacity 3 months post dose 2 (±1 month). METHODS: COVID-19 neutralization efficiency was measured by calculating the half maximal inhibitory dilution (ID50) using a high-throughput protein-based neutralization assay for Ancestral (Wuhan), Delta and Omicron (BA.1) spike variants. Univariable and multivariable quantile regression were used to compare COVID-19-specific antibody neutralization capacity by HIV status. RESULTS: Neutralization assays were performed on 256 PWH and 256 controls based on specimen availability at the timepoint of interest, having received two vaccines and known date of vaccination. There was a significant interaction between HIV status and previous COVID-19 infection status in median ID50. There were no differences in median ID50 for HIV+ vs. HIV-negative persons without past COVID-19 infection. For participants with past COVID-19 infection, median ICD50 was significantly higher in controls than in PWH for ancestral SARS-CoV-2 and Omicron variants, with a trend for the Delta variant in the same direction. CONCLUSION: Vaccine-induced SARS-CoV-2 neutralization capacity was similar between PWH vs. HIV-negative persons without past COVID-19 infection, demonstrating favourable humoral-mediated immunogenicity. Both HIV+ and HIV-negative persons demonstrated hybrid immunity. TRIAL REGISTRATION: clinicaltrials.gov NCT04894448.

Cross-sectional Characterization of SARS-CoV-2 Antibody Levels and Decay Rates Following Infection of Unvaccinated Elderly Individuals.

Background: SARS-CoV-2 infections have disproportionally burdened elderly populations with excessive mortality. While several contributing factors exists, questions remain about the quality and duration of humoral antibody-mediated responses resulting from infections in unvaccinated elderly individuals. Methods: Residual serum/plasma samples were collected from individuals undergoing routine SARS-CoV-2 polymerase chain reaction testing in a community laboratory in Canada. The samples were collected in 2020, before vaccines became available. IgG, IgA, and IgM antibodies against SARS-CoV-2 nucleocapsid, trimeric spike, and its receptor-binding domain were quantified via a high-throughput chemiluminescent enzyme-linked immunosorbent assay. Neutralization efficiency was also quantified through a surrogate high-throughput protein-based neutralization assay. Results: This study analyzed SARS-CoV-2 antibody levels in a large cross-sectional cohort (N = 739), enriched for elderly individuals (median age, 82 years; 75% >65 years old), where 72% of samples tested positive for SARS-CoV-2 by polymerase chain reaction. The age group ≥90 years had higher levels of antibodies than that <65 years. Neutralization efficiency showed an age-dependent trend, where older persons had higher levels of neutralizing antibodies. Antibodies targeting the nucleocapsid had the fastest decline. IgG antibodies targeting the receptor-binding domain remained stable over time, potentially explaining the lack of neutralization decay observed in this cohort. Conclusions: Despite older individuals having the highest levels of antibodies postinfection, they are the cohort in which antibody decay was the fastest. Until a better understanding of correlates of protection is acquired, along with the protective role of nonneutralizing antibodies, booster vaccinations remain important in this demographic.

Machine Learning Approaches to Classify Self-Reported Rheumatoid Arthritis Health Scores Using Activity Tracker Data: Longitudinal Observational Study.

BACKGROUND: The increasing use of activity trackers in mobile health studies to passively collect physical data has shown promise in lessening participation burden to provide actively contributed patient-reported outcome (PRO) information. OBJECTIVE: The aim of this study was to develop machine learning models to classify and predict PRO scores using Fitbit data from a cohort of patients with rheumatoid arthritis. METHODS: Two different models were built to classify PRO scores: a random forest classifier model that treated each week of observations independently when making weekly predictions of PRO scores, and a hidden Markov model that additionally took correlations between successive weeks into account. Analyses compared model evaluation metrics for (1) a binary task of distinguishing a normal PRO score from a severe PRO score and (2) a multiclass task of classifying a PRO score state for a given week. RESULTS: For both the binary and multiclass tasks, the hidden Markov model significantly (P<.05) outperformed the random forest model for all PRO scores, and the highest area under the curve, Pearson correlation coefficient, and Cohen κ coefficient were 0.750, 0.479, and 0.471, respectively. CONCLUSIONS: While further validation of our results and evaluation in a real-world setting remains, this study demonstrates the ability of physical activity tracker data to classify health status over time in patients with rheumatoid arthritis and enables the possibility of scheduling preventive clinical interventions as needed. If patient outcomes can be monitored in real time, there is potential to improve clinical care for patients with other chronic conditions.

Federated Learning with Research Prototypes: Application to Multi-Center MRI-based Detection of Prostate Cancer with Diverse Histopathology.

RATIONALE AND OBJECTIVES: Early prostate cancer detection and staging from MRI is extremely challenging for both radiologists and deep learning algorithms, but the potential to learn from large and diverse datasets remains a promising avenue to increase their performance within and across institutions. To enable this for prototype-stage algorithms, where the majority of existing research remains, we introduce a flexible federated learning framework for cross-site training, validation, and evaluation of custom deep learning prostate cancer detection algorithms. MATERIALS AND METHODS: We introduce an abstraction of prostate cancer groundtruth that represents diverse annotation and histopathology data. We maximize use of this groundtruth if and when they are available using UCNet, a custom 3D UNet that enables simultaneous supervision of pixel-wise, region-wise, and gland-wise classification. We leverage these modules to perform cross-site federated training using 1400+ heterogeneous multi-parameteric prostate MRI exams from two University hospitals. RESULTS: We observe a positive result, with significant improvements in cross-site generalization performance with negligible intra-site performance degradation for both lesion segmentation and per-lesion binary classification of clinically-significant prostate cancer. Cross-site lesion segmentation performance intersection-over-union (IoU) improved by 100%, while cross-site lesion classification performance overall accuracy improved by 9.5-14.8%, depending on the optimal checkpoint selected by each site. CONCLUSION: Federated learning can improve the generalization performance of prostate cancer detection models across institutions while protecting patient health information and institution-specific code and data. However, even more data and participating institutions are likely required to improve the absolute performance of prostate cancer classification models. To enable adoption of federated learning with limited re-engineering of federated components, we open-source our FLtools system at https://federated.ucsf.edu, including examples that can be easily adapted to other medical imaging deep learning projects.

Multi-Parametric Radiomic Model to Predict 1p/19q Co-Deletion in Patients with IDH-1 Mutant Glioma: Added Value to the T2-FLAIR Mismatch Sign.

PURPOSE: The T2-FLAIR mismatch sign has shown promise in determining IDH mutant 1p/19q non-co-deleted gliomas with a high specificity and modest sensitivity. To develop a multi-parametric radiomic model using MRI to predict 1p/19q co-deletion status in patients with newly diagnosed IDH1 mutant glioma and to perform a comparative analysis to T2-FLAIR mismatch sign+. METHODS: In this retrospective study, patients with diagnosis of IDH1 mutant gliomas with known 1p/19q status who had preoperative MRI were included. T2-FLAIR mismatch was evaluated independently by two board-certified neuroradiologists. Texture features were extracted from glioma segmentation of FLAIR images. eXtremeGradient Boosting (XGboost) classifiers were used for model development. Leave-one-out-cross-validation (LOOCV) and external validation performances were reported for both the training and external validation sets. RESULTS: A total of 103 patients were included for model development and 18 patients for external testing validation. The diagnostic performance (sensitivity/specificity/accuracy) in the determination of the 1p/19q co-deletion status was 59%/83%/67% (training) and 62.5%/70.0%/66.3% (testing) for the T2-FLAIR mismatch sign. This was significantly improved (p = 0.04) using the radiomics model to 77.9%/82.8%/80.3% (training) and 87.5%/89.9%/88.8% (testing), respectively. The addition of radiomics as a computer-assisted tool resulted in significant (p = 0.02) improvement in the performance of the neuroradiologist with 13 additional corrected cases in comparison to just using the T2-FLAIR mismatch sign. CONCLUSION: The proposed radiomic model provides much needed sensitivity to the highly specific T2-FLAIR mismatch sign in the determination of the 1p/19q non-co-deletion status and improves the overall diagnostic performance of neuroradiologists when used as an assistive tool.

A Multitask Approach for Automated Detection and Segmentation of Thyroid Nodules in Ultrasound Images.

An increase in the incidence and diagnosis of thyroid nodules and thyroid cancer underscores the need for a better approach to nodule detection and risk stratification in ultrasound (US) images that can reduce healthcare costs, patient discomfort, and unnecessary invasive procedures. However, variability in ultrasound technique and interpretation makes the diagnostic process partially subjective. Therefore, an automated approach that detects and segments nodules could improve performance on downstream tasks, such as risk stratification.Current deep learning architectures for segmentation are typically semi-automated because they are evaluated solely on images known to have nodules and do not assess ability to identify suspicious images. However, the proposed multitask approach both detects suspicious images and segments potential nodules; this allows for a clinically translatable model that aptly parallels the workflow for thyroid nodule assessment. The multitask approach is centered on an anomaly detection (AD) module that can be integrated with any U-Net architecture variant to improve image-level nodule detection. Ultrasound studies were acquired from 280 patients at UCLA Health, totaling 9,888 images, and annotated by collaborating radiologists. Of the evaluated models, a multi-scale UNet (MSUNet) with AD achieved the highest F1 score of 0.829 and image-wide Dice similarity coefficient of 0.782 on our hold-out test set. Furthermore, models were evaluated on two external validations datasets to demonstrate generalizability and robustness to data variability. Ultimately, the proposed architecture is an automated multitask method that expands on previous methods by successfully both detecting and segmenting nodules in ultrasound.

Institutional Strategies to Maintain and Grow Imaging Research During the COVID-19 Pandemic.

Understanding imaging research experiences, challenges, and strategies for academic radiology departments during and after COVID-19 is critical to prepare for future disruptive events. We summarize key insights and programmatic initiatives at major academic hospitals across the world, based on literature review and meetings of the Radiological Society of North America Vice Chairs of Research (RSNA VCR) group. Through expert discussion and case studies, we provide suggested guidelines to maintain and grow radiology research in the postpandemic era.