Canadian radiology: 2024 update.

Radiology in Canada is advancing through innovations in clinical practices and research methodologies. Recent developments focus on refining evidence-based practice guidelines, exploring innovative imaging techniques and enhancing diagnostic processes through artificial intelligence. Within the global radiology community, Canadian institutions play an important role by engaging in international collaborations, such as with the American College of Radiology to refine implementation of the Ovarian-Adnexal Reporting and Data System for ultrasound and magnetic resonance imaging. Additionally, researchers have participated in multidisciplinary collaborations to evaluate the performance of artificial intelligence-driven diagnostic tools for chronic liver disease and pediatric brain tumors. Beyond clinical radiology, efforts extend to addressing gender disparities in the field, improving educational practices, and enhancing the environmental sustainability of radiology departments. These advancements highlight Canada's role in the global radiology community, showcasing a commitment to improving patient outcomes and advancing the field through research and innovation. This update underscores the importance of continued collaboration and innovation to address emerging challenges and further enhance the quality and efficacy of radiology practices worldwide.

Evaluation of a BERT Natural Language Processing Model for Automating CT and MRI Triage and Protocol Selection.

Purpose: To evaluate the accuracy of a Bidirectional Encoder Representations for Transformers (BERT) Natural Language Processing (NLP) model for automating triage and protocol selection of cross-sectional image requisitions. Methods: A retrospective study was completed using 222 392 CT and MRI studies from a single Canadian university hospital database (January 2018-September 2022). Three hundred unique protocols (116 CT and 184 MRI) were included. A BERT model was trained, validated, and tested using an 80%-10%-10% stratified split. Naive Bayes (NB) and Support Vector Machine (SVM) machine learning models were used as comparators. Models were assessed using F1 score, precision, recall, and area under the receiver operating characteristic curve (AUROC). The BERT model was also assessed for multi-class protocol suggestion and subgroups based on referral location, modality, and imaging section. Results: BERT was superior to SVM for protocol selection (F1 score: BERT-0.901 vs SVM-0.881). However, was not significantly different from SVM for triage prediction (F1 score: BERT-0.844 vs SVM-0.845). Both models outperformed NB for protocol and triage. BERT had superior performance on minority classes compared to SVM and NB. For multiclass prediction, BERT accuracy was up to 0.991 for top-5 protocol suggestion, and 0.981 for top-2 triage suggestion. Emergency department patients had the highest F1 scores for both protocol (0.957) and triage (0.986), compared to inpatients and outpatients. Conclusion: The BERT NLP model demonstrated strong performance in automating the triage and protocol selection of radiology studies, showing potential to enhance radiologist workflows. These findings suggest the feasibility of using advanced NLP models to streamline radiology operations.

Applications of Artificial Intelligence in Acute Abdominal Imaging.

Artificial intelligence (AI) is a rapidly growing field with significant implications for radiology. Acute abdominal pain is a common clinical presentation that can range from benign conditions to life-threatening emergencies. The critical nature of these situations renders emergent abdominal imaging an ideal candidate for AI applications. CT, radiographs, and ultrasound are the most common modalities for imaging evaluation of these patients. For each modality, numerous studies have assessed the performance of AI models for detecting common pathologies, such as appendicitis, bowel obstruction, and cholecystitis. The capabilities of these models range from simple classification to detailed severity assessment. This narrative review explores the evolution, trends, and challenges in AI applications for evaluating acute abdominal pathologies. We review implementations of AI for non-traumatic and traumatic abdominal pathologies, with discussion of potential clinical impact, challenges, and future directions for the technology.

Machine and deep learning models for accurate detection of ischemia and scar with myocardial blood flow positron emission tomography imaging.

BACKGROUND: Quantification of myocardial blood flow (MBF) is used for the noninvasive diagnosis of patients with coronary artery disease (CAD). This study compared traditional statistics, machine learning, and deep learning techniques in their ability to diagnose disease using only the rest and stress MBF values. METHODS: This study included 3245 rest and stress rubidium-82 positron emission tomography (PET) studies and matching diagnostic labels from perfusion reports. Standard logistic regression, lasso logistic regression, support vector machine, random forest, multilayer perceptron, and dense U-Net were compared for per-patient detection and per-vessel localization of scars and ischemia. RESULTS: Receiver-operator characteristic area under the curve (AUC) of machine learning models was significantly higher than those of traditional statistics models for per-patient detection of disease (0.92-0.95 vs. 0.87) but not for per-vessel localization of ischemia or scar. Random forest showed the highest AUC = 0.95 among the different models compared. On the final hold-out set for generalizability, random forest showed an AUC of 0.92 for detection and 0.89 for localization of perfusion abnormalities. CONCLUSIONS: For per-vessel localization, simple models trained on segmental data performed similarly to a convolutional neural network trained on polar-map data, highlighting the need to justify the use of complex predictive algorithms through comparison with simpler methods.

Artificial Intelligence Chatbots' Understanding of the Risks and Benefits of Computed Tomography and Magnetic Resonance Imaging Scenarios.

PURPOSE: Patients may seek online information to better understand medical imaging procedures. The purpose of this study was to assess the accuracy of information provided by 2 popular artificial intelligence (AI) chatbots pertaining to common imaging scenarios' risks, benefits, and alternatives. METHODS: Fourteen imaging-related scenarios pertaining to computed tomography (CT) or magnetic resonance imaging (MRI) were used. Factors including the use of intravenous contrast, the presence of renal disease, and whether the patient was pregnant were included in the analysis. For each scenario, 3 prompts for outlining the (1) risks, (2) benefits, and (3) alternative imaging choices or potential implications of not using contrast were inputted into ChatGPT and Bard. A grading rubric and a 5-point Likert scale was used by 2 independent reviewers to grade responses. Prompt variability and chatbot context dependency were also assessed. RESULTS: ChatGPT's performance was superior to Bard's in accurately responding to prompts per Likert grading (4.36 ± 0.63 vs 3.25 ± 1.03 seconds, P < .0001). There was substantial agreement between independent reviewer grading for ChatGPT (κ = 0.621) and Bard (κ = 0.684). Response text length was not statistically different between ChatGPT and Bard (2087 ± 256 characters vs 2162 ± 369 characters, P = .24). Response time was longer for ChatGPT (34 ± 2 vs 8 ± 1 seconds, P < .0001). CONCLUSIONS: ChatGPT performed superior to Bard at outlining risks, benefits, and alternatives to common imaging scenarios. Generally, context dependency and prompt variability did not change chatbot response content. Due to the lack of detailed scientific reasoning and inability to provide patient-specific information, both AI chatbots have limitations as a patient information resource.

Quantum critical dynamics in a 5,000-qubit programmable spin glass.

Experiments on disordered alloys1-3 suggest that spin glasses can be brought into low-energy states faster by annealing quantum fluctuations than by conventional thermal annealing. Owing to the importance of spin glasses as a paradigmatic computational testbed, reproducing this phenomenon in a programmable system has remained a central challenge in quantum optimization4-13. Here we achieve this goal by realizing quantum-critical spin-glass dynamics on thousands of qubits with a superconducting quantum annealer. We first demonstrate quantitative agreement between quantum annealing and time evolution of the Schrödinger equation in small spin glasses. We then measure dynamics in three-dimensional spin glasses on thousands of qubits, for which classical simulation of many-body quantum dynamics is intractable. We extract critical exponents that clearly distinguish quantum annealing from the slower stochastic dynamics of analogous Monte Carlo algorithms, providing both theoretical and experimental support for large-scale quantum simulation and a scaling advantage in energy optimization.

Evaluation of a deep learning method for the automated detection of supraspinatus tears on MRI.

OBJECTIVE: To evaluate if deep learning is a feasible approach for automated detection of supraspinatus tears on MRI. MATERIALS AND METHODS: A total of 200 shoulder MRI studies performed between 2015 and 2019 were retrospectively obtained from our institutional database using a balanced random sampling of studies containing a full-thickness tear, partial-thickness tear, or intact supraspinatus tendon. A 3-stage pipeline was developed comprised of a slice selection network based on a pre-trained residual neural network (ResNet); a segmentation network based on an encoder-decoder network (U-Net); and a custom multi-input convolutional neural network (CNN) classifier. Binary reference labels were created following review of radiologist reports and images by a radiology fellow and consensus validation by two musculoskeletal radiologists. Twenty percent of the data was reserved as a holdout test set with the remaining 80% used for training and optimization under a fivefold cross-validation strategy. Classification and segmentation accuracy were evaluated using area under the receiver operating characteristic curve (AUROC) and Dice similarity coefficient, respectively. Baseline characteristics in correctly versus incorrectly classified cases were compared using independent sample t-test and chi-squared. RESULTS: Test sensitivity and specificity of the classifier at the optimal Youden's index were 85.0% (95% CI: 62.1-96.8%) and 85.0% (95% CI: 62.1-96.8%), respectively. AUROC was 0.943 (95% CI: 0.820-0.991). Dice segmentation accuracy was 0.814 (95% CI: 0.805-0.826). There was no significant difference in AUROC between 1.5 T and 3.0 T studies. Sub-analysis showed superior sensitivity on full-thickness (100%) versus partial-thickness (72.5%) subgroups. DATA CONCLUSION: Deep learning is a feasible approach to detect supraspinatus tears on MRI.

Prognostic utility of longitudinal quantification of PET myocardial blood flow early post heart transplantation.

BACKGROUND: Myocardial blood flow (MBF) quantification by Rubidium-82 positron emission tomography (PET) has shown promise for cardiac allograft vasculopathy (CAV) surveillance and risk stratification post heart transplantation. The objective was to determine the prognostic value of serial PET performed early post transplantation. METHODS AND RESULT: Heart transplant (HT) recipients at the University of Ottawa Heart Institute with 2 PET examinations (PET1 = baseline, PET2 = follow-up) within 6 years of transplant were included in the study. Evaluation of PET flow quantification included stress MBF, coronary vascular resistance (CVR), and myocardial flow reserve (MFR). The primary composite outcome was all-cause death, re-transplant, myocardial infarction, revascularization, allograft dysfunction, cardiac allograft vasculopathy (CAV), or heart failure hospitalization. A total of 121 patients were evaluated (79% male, mean age 56 ± 11 years) with consecutive scans performed at mean 1.4 ± 0.7 and 2.6 ± 1.0 years post HT for PET1 and PET2, respectively. Over a mean follow-up of 3.0 (IQR 1.8, 4.6) years, 26 (22%) patients developed the primary outcome: 1 death, 11 new or progressive angiographic CAV, 2 percutaneous coronary interventions, 12 allograft dysfunction. Unadjusted Cox analysis showed a significant reduction in event-free survival in patients with PET1 stress MBF < 2.1 (HR: 2.43, 95% CI 1.11-5.29 P = 0.047) and persistent abnormal PET1 to PET2 CVR > 76 (HR: 2.19, 95% CI 0.87-5.51 P = 0.045). There was no association between MFR and outcomes. CONCLUSION: Low-stress MBF and persistent increased CVR on serial PET imaging early post HT are associated with adverse cardiovascular outcomes. Early post-transplant and longitudinal assessment by PET may identify at-risk patients for increased surveillance post HT.

Scaling advantage over path-integral Monte Carlo in quantum simulation of geometrically frustrated magnets.

The promise of quantum computing lies in harnessing programmable quantum devices for practical applications such as efficient simulation of quantum materials and condensed matter systems. One important task is the simulation of geometrically frustrated magnets in which topological phenomena can emerge from competition between quantum and thermal fluctuations. Here we report on experimental observations of equilibration in such simulations, measured on up to 1440 qubits with microsecond resolution. By initializing the system in a state with topological obstruction, we observe quantum annealing (QA) equilibration timescales in excess of one microsecond. Measurements indicate a dynamical advantage in the quantum simulation compared with spatially local update dynamics of path-integral Monte Carlo (PIMC). The advantage increases with both system size and inverse temperature, exceeding a million-fold speedup over an efficient CPU implementation. PIMC is a leading classical method for such simulations, and a scaling advantage of this type was recently shown to be impossible in certain restricted settings. This is therefore an important piece of experimental evidence that PIMC does not simulate QA dynamics even for sign-problem-free Hamiltonians, and that near-term quantum devices can be used to accelerate computational tasks of practical relevance.

Internal validation of myocardial flow reserve PET imaging using stress/rest myocardial activity ratios with Rb-82 and N-13-ammonia.

BACKGROUND: Myocardial flow reserve (MFR) measurement provides incremental diagnostic and prognostic information. The objective of the current study was to investigate the application of a simplified model for the estimation of MFR using only the stress/rest myocardial activity ratio (MAR) in patients undergoing rest-stress cardiac PET MPI. METHODS AND RESULTS: Rest and dipyridamole stress dynamic PET imaging was performed in consecutive patients using 82Rb or 13NH3 (n = 250 each). Reference standard MFR was quantified using a standard one-tissue compartment model. Stress/rest myocardial activity ratio (MAR) was calculated using the LV-mean activity from 2 to 6 minutes post-injection. Simplified estimates of MFR (MFREST) were then calculated using an inverse power function. For 13NH3, there was good correlation between MFR and MFREST values (R = 0.63), with similar results for 82Rb (R = 0.73). There was no bias in the MFREST values with either tracer. The overall diagnostic performance of MFREST for detection of MFR < 2 was good with ROC area under the curve (AUC) = 83.2 ± 1.2% for 13NH3 and AUC = 90.4 ± 0.7% for 82Rb. CONCLUSION: MFR was estimated with good accuracy using 82Rb and 13NH3 with a simplified method that relies only on stress/rest activity ratios. This novel approach does not require dynamic imaging or tracer kinetic modeling. It may be useful for routine quality assurance of PET MFR measurements, or in scanners where full dynamic imaging and tracer kinetic modeling is not feasible for technical or logistical reasons.

Publication bias in diagnostic imaging: conference abstracts with positive conclusions are more likely to be published.

OBJECTIVE: To evaluate whether imaging diagnostic test accuracy conference abstracts with positive conclusions or titles are more likely to reach full-text publication than those with negative (or neutral) conclusions or titles. METHODS: Diagnostic accuracy research abstracts were included if they were presented at the 2011 or 2012 Radiological Society of North America conference. Full-text publication status at 5 years post conference abstract submission was determined. Conclusion and title positivity of conference abstracts were extracted, as well as potential confounding factors. The associations of conclusion and title positivity with publication status at 5 years post conference abstract submission were assessed using a multivariable logistic regression model. Conditional odds ratios were calculated to express the strength of associations, adjusting for the confounders. RESULTS: In total, 282/400 (71%) of included conference abstracts reached full-text publication. A total of 246 out of 337 (74%) conference abstracts with positive conclusions resulted in full-text publications, compared with 26/48 (54%) with neutral conclusions and 5/15 (33%) with negative conclusions. In multivariable logistic regression, conclusion positivity was significantly associated with full-text publication (odds ratio 3.6; 95% CI 1.9-6.7 for conference abstracts with positive conclusions, compared with those with non-positive conclusions); this did not apply to title positivity (odds ratio 1.2; 95% CI 0.47-3.0). CONCLUSION: Imaging conference abstracts with positive conclusions were more likely to be published as full-text articles. Title positivity was not associated with publication. This preferential publication pattern may lead to an overrepresentation of positive studies in the literature. An overrepresentation of positive studies may contribute to inflated estimates of test accuracy and has the potential to adversely influence patient care. KEY POINTS: • Imaging diagnostic test accuracy conference abstracts with positive conclusions were more likely to be reported as full-text articles than those with non-positive conclusions. • The majority (75%) of imaging diagnostic test accuracy conference abstracts with positive conclusions were published, compared with only 53% and 33% with neutral and negative conclusions, respectively. • Conclusion positivity remained associated with the full-text publication of conference abstracts when controlling for multiple potential confounding variables.

Observation of topological phenomena in a programmable lattice of 1,800 qubits.

The work of Berezinskii, Kosterlitz and Thouless in the 1970s1,2 revealed exotic phases of matter governed by the topological properties of low-dimensional materials such as thin films of superfluids and superconductors. A hallmark of this phenomenon is the appearance and interaction of vortices and antivortices in an angular degree of freedom-typified by the classical XY model-owing to thermal fluctuations. In the two-dimensional Ising model this angular degree of freedom is absent in the classical case, but with the addition of a transverse field it can emerge from the interplay between frustration and quantum fluctuations. Consequently, a Kosterlitz-Thouless phase transition has been predicted in the quantum system-the two-dimensional transverse-field Ising model-by theory and simulation3-5. Here we demonstrate a large-scale quantum simulation of this phenomenon in a network of 1,800 in situ programmable superconducting niobium flux qubits whose pairwise couplings are arranged in a fully frustrated square-octagonal lattice. Essential to the critical behaviour, we observe the emergence of a complex order parameter with continuous rotational symmetry, and the onset of quasi-long-range order as the system approaches a critical temperature. We describe and use a simple approach to statistical estimation with an annealing-based quantum processor that performs Monte Carlo sampling in a chain of reverse quantum annealing protocols. Observations are consistent with classical simulations across a range of Hamiltonian parameters. We anticipate that our approach of using a quantum processor as a programmable magnetic lattice will find widespread use in the simulation and development of exotic materials.

RNA-MoIP: prediction of RNA secondary structure and local 3D motifs from sequence data.

RNA structures are hierarchically organized. The secondary structure is articulated around sophisticated local three-dimensional (3D) motifs shaping the full 3D architecture of the molecule. Recent contributions have identified and organized recurrent local 3D motifs, but applications of this knowledge for predictive purposes is still in its infancy. We recently developed a computational framework, named RNA-MoIP, to reconcile RNA secondary structure and local 3D motif information available in databases. In this paper, we introduce a web service using our software for predicting RNA hybrid 2D-3D structures from sequence data only. Optionally, it can be used for (i) local 3D motif prediction or (ii) the refinement of user-defined secondary structures. Importantly, our web server automatically generates a script for the MC-Sym software, which can be immediately used to quickly predict all-atom RNA 3D models. The web server is available at http://rnamoip.cs.mcgill.ca.

Distinct cellular states determine calcium signaling response.

The heterogeneity in mammalian cells signaling response is largely a result of pre-existing cell-to-cell variability. It is unknown whether cell-to-cell variability rises from biochemical stochastic fluctuations or distinct cellular states. Here, we utilize calcium response to adenosine trisphosphate as a model for investigating the structure of heterogeneity within a population of cells and analyze whether distinct cellular response states coexist. We use a functional definition of cellular state that is based on a mechanistic dynamical systems model of calcium signaling. Using Bayesian parameter inference, we obtain high confidence parameter value distributions for several hundred cells, each fitted individually. Clustering the inferred parameter distributions revealed three major distinct cellular states within the population. The existence of distinct cellular states raises the possibility that the observed variability in response is a result of structured heterogeneity between cells. The inferred parameter distribution predicts, and experiments confirm that variability in IP3R response explains the majority of calcium heterogeneity. Our work shows how mechanistic models and single-cell parameter fitting can uncover hidden population structure and demonstrate the need for parameter inference at the single-cell level.

Signal Transduction at the Single-Cell Level: Approaches to Study the Dynamic Nature of Signaling Networks.

Signal transduction, or how cells interpret and react to external events, is a fundamental aspect of cellular function. Traditional study of signal transduction pathways involves mapping cellular signaling pathways at the population level. However, population-averaged readouts do not adequately illuminate the complex dynamics and heterogeneous responses found at the single-cell level. Recent technological advances that observe cellular response, computationally model signaling pathways, and experimentally manipulate cells now enable studying signal transduction at the single-cell level. These studies will enable deeper insights into the dynamic nature of signaling networks.

Systems biology. Accurate information transmission through dynamic biochemical signaling networks.

Stochasticity inherent to biochemical reactions (intrinsic noise) and variability in cellular states (extrinsic noise) degrade information transmitted through signaling networks. We analyzed the ability of temporal signal modulation--that is, dynamics--to reduce noise-induced information loss. In the extracellular signal-regulated kinase (ERK), calcium (Ca(2+)), and nuclear factor kappa-B (NF-κB) pathways, response dynamics resulted in significantly greater information transmission capacities compared to nondynamic responses. Theoretical analysis demonstrated that signaling dynamics has a key role in overcoming extrinsic noise. Experimental measurements of information transmission in the ERK network under varying signal-to-noise levels confirmed our predictions and showed that signaling dynamics mitigate, and can potentially eliminate, extrinsic noise-induced information loss. By curbing the information-degrading effects of cell-to-cell variability, dynamic responses substantially increase the accuracy of biochemical signaling networks.