Availability of ultraportable chest x-ray (CXR) and advancements in artificial intelligence (AI)-enabled CXR interpretation are promising developments in tuberculosis (TB) active case finding (ACF) but costing and cost-effectiveness analyses are limited. We provide implementation cost and cost-effectiveness estimates of different screening algorithms using symptoms, CXR and AI in Nigeria. People 15 years and older were screened for TB symptoms and offered a CXR with AI-enabled interpretation using qXR v3 (Qure.ai) at lung health camps. Sputum samples were tested on Xpert MTB/RIF for individuals reporting symptoms or with qXR abnormality scores ≥0.30. We conducted a retrospective costing using a combination of top-down and bottom-up approaches while utilizing itemized expense data from a health system perspective. We estimated costs in five screening scenarios: abnormality score ≥0.30 and ≥0.50; cough ≥ 2 weeks; any symptom; abnormality score ≥0.30 or any symptom. We calculated total implementation costs, cost per bacteriologically-confirmed case detected, and assessed cost-effectiveness using incremental cost-effectiveness ratio (ICER) as additional cost per additional case. Overall, 3205 people with presumptive TB were identified, 1021 were tested, and 85 people with bacteriologically-confirmed TB were detected. Abnormality ≥ 0.30 or any symptom (US$65704) had the highest costs while cough ≥ 2 weeks was the lowest (US$40740). The cost per case was US$1198 for cough ≥ 2 weeks, and lowest for any symptom (US$635). Compared to baseline strategy of cough ≥ 2 weeks, the ICER for any symptom was US$191 per additional case detected and US$ 2096 for Abnormality ≥0.30 OR any symptom algorithm. Using CXR and AI had lower cost per case detected than any symptom screening criteria when asymptomatic TB was higher than 30% of all bacteriologically-confirmed TB detected. Compared to traditional symptom screening, using CXR and AI in combination with symptoms detects more cases at lower cost per case detected and is cost-effective. TB programs should explore adoption of CXR and AI for screening in ACF.

Pulmonary embolism (PE) is the third most fatal cardiovascular disease in the United States. Currently, Computed Tomography Pulmonary Angiography (CTPA) serves as diagnostic gold standard for detecting PE. However, its efficacy is limited by factors such as contrast bolus timing, physician-dependent diagnostic accuracy, and time taken for scan interpretation. To address these limitations, we propose an AI-based PE triaging model (AID-PE) designed to predict the presence and key characteristics of PE on CTPA. This model aims to enhance diagnostic accuracy, efficiency, and the speed of PE identification. We trained AID-PE on the RSNA-STR PE CT (RSPECT) Dataset, N = 7279 and subsequently tested it on an in-house dataset (n = 106). We evaluated efficiency in a separate dataset (D₄, n = 200) by comparing the time from scan to report in standard PE detection workflow versus AID-PE. A comparative analysis showed that AID-PE had an AUC/accuracy of 0.95/0.88. In contrast, a Convolutional Neural Network (CNN) classifier and a CNN-Long Short-Term Memory (LSTM) network without an attention module had an AUC/accuracy of 0.5/0.74 and 0.88/0.65, respectively. Our model achieved AUCs of 0.82 and 0.95 for detecting PE on the validation dataset and the independent test set, respectively. On D₄, AID-PE took an average of 1.32 s to screen for PE across 148 CTPA studies, compared to an average of 40 min in contemporary workflow. AID-PE outperformed a baseline CNN classifier and a single-stage CNN-LSTM network without an attention module. Additionally, its efficiency is comparable to the current radiological workflow.

Fully-supervised deep neural networks have achieved remarkable progress in medical image segmentation, yet they heavily rely on extensive manually labeled data and exhibit inflexibility for unseen tasks. Few-shot segmentation (FSS) addresses these issues by predicting unseen classes from a few labeled support examples. However, most existing FSS models struggle to generalize to diverse target tasks distinct from training domains. Furthermore, designing promising network architectures for such tasks is expertise-intensive and laborious. In this paper, we introduce MOE-FewSeg, a novel automatic design method for FSS architectures. Specifically, we construct a U-shaped encoder-decoder search space that incorporates capabilities for information interaction and feature selection, thereby enabling architectures to leverage prior knowledge from publicly available datasets across diverse domains for improved prediction of various target tasks. Given the potential conflicts among disparate target tasks, we formulate the multi-task problem as a multi-objective optimization problem. We employ a multi-objective genetic algorithm to identify the Pareto-optimal architectures for these target tasks within this search space. Furthermore, to mitigate the impact of noisy labels due to dataset quality variations, we propose a noise-robust loss function named NRL, which encourages the model to de-emphasize larger loss values. Empirical results demonstrate that MOE-FewSeg outperforms manually designed architectures and other related approaches.

Hip fractures are a major orthopedic problem, especially in the elderly population. Hip fractures are usually diagnosed by clinical evaluation and imaging, especially X-rays. In recent years, new approaches to fracture detection have emerged with the use of artificial intelligence (AI) and deep learning techniques in medical imaging. In this study, we aimed to evaluate the diagnostic performance of ChatGPT-4o, an artificial intelligence model, in diagnosing hip fractures. A total of 200 anteroposterior pelvic X-ray images were retrospectively analyzed. Half of the images belonged to patients with surgically confirmed hip fractures, including both displaced and non-displaced types, while the other half represented patients with soft tissue trauma and no fractures. Each image was evaluated by ChatGPT-4o through a standardized prompt, and its predictions (fracture vs. no fracture) were compared against the gold standard diagnoses. Diagnostic performance metrics such as sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), receiver operating characteristic (ROC) curve, Cohen's kappa, and F1 score were calculated. ChatGPT-4o demonstrated an overall accuracy of 82.5% in detecting hip fractures on pelvic radiographs, with a sensitivity of 78.0% and specificity of 87.0%. PPVs and NPVs were 85.7% and 79.8%, respectively. The area under the ROC curve (AUC) was 0.825, indicating good discriminative performance. Among 22 false-negative cases, 68.2% were non-displaced fractures, suggesting the model had greater difficulty identifying subtle radiographic findings. Cohen's kappa coefficient was 0.65, showing substantial agreement with actual diagnoses. Chi-square analysis revealed a strong correlation (χ² = 82.59, <i>P</i> < 0.001), while McNemar's test (<i>P</i> = 0.176) showed no significant asymmetry in error distribution. ChatGPT-4o shows promising accuracy in identifying hip fractures on pelvic X-rays, especially when fractures are displaced. However, its sensitivity drops significantly for non-displaced fractures, leading to many false negatives. This highlights the need for caution when interpreting negative AI results, particularly when clinical suspicion remains high. While not a replacement for expert assessment, ChatGPT-4o may assist in settings with limited specialist access.

Interpretable deep-learning models have received widespread attention in the field of image recognition. However, owing to the coexistence of medical-image categories and the challenge of identifying subtle decision-making regions, many proposed interpretable deep-learning models suffer from insufficient accuracy and interpretability in diagnosing images of medical diseases. Therefore, this study proposed a feature-driven inference network (FeaInfNet) that incorporates a feature-based network reasoning structure. Specifically, local feature masks (LFM) were developed to extract feature vectors, thereby providing global information for these vectors and enhancing the expressive ability of FeaInfNet. Second, FeaInfNet compares the similarity of the feature vector corresponding to each subregion image patch with the disease and normal prototype templates that may appear in the region. It then combines the comparison of each subregion when making the final diagnosis. This strategy simulates the diagnosis process of doctors, making the model interpretable during the reasoning process, while avoiding misleading results caused by the participation of normal areas during reasoning. Finally, we proposed adaptive dynamic masks (Adaptive-DM) to interpret feature vectors and prototypes into human-understandable image patches to provide an accurate visual interpretation. Extensive experiments on multiple publicly available medical datasets, including RSNA, iChallenge-PM, COVID-19, ChinaCXRSet, MontgomerySet, and CBIS-DDSM, demonstrated that our method achieves state-of-the-art classification accuracy and interpretability compared with baseline methods in the diagnosis of medical images. Additional ablation studies were performed to verify the effectiveness of each component.

Brain disorder diagnosis via resting-state functional magnetic resonance imaging (rs-fMRI) is usually limited due to the complex imaging features and sample size. For brain disorder diagnosis, the graph convolutional network (GCN) has achieved remarkable success by capturing interactions between individuals and the population. However, there are mainly three limitations: 1) The previous GCN approaches consider the non-imaging information in edge construction but ignore the sensitivity differences of features to non-imaging information. 2) The previous GCN approaches solely focus on establishing interactions between subjects (i.e., individuals and the population), disregarding the essential relationship between features. 3) Multisite data increase the sample size to help classifier training, but the inter-site heterogeneity limits the performance to some extent. This paper proposes a knowledge-aware multisite adaptive graph Transformer to address the above problems. First, we evaluate the sensitivity of features to each piece of non-imaging information, and then construct feature-sensitive and feature-insensitive subgraphs. Second, after fusing the above subgraphs, we integrate a Transformer module to capture the intrinsic relationship between features. Third, we design a domain adaptive GCN using multiple loss function terms to relieve data heterogeneity and to produce the final classification results. Last, the proposed framework is validated on two brain disorder diagnostic tasks. Experimental results show that the proposed framework can achieve state-of-the-art performance.

In Computed Tomography (CT) imaging, the ring artifacts caused by the inconsistent detector response can significantly degrade the reconstructed images, having negative impacts on the subsequent applications. The new generation of CT systems based on photon-counting detectors are affected by ring artifacts more severely. The flexibility and variety of detector responses make it difficult to build a well-defined model to characterize the ring artifacts. In this context, this study proposes the global dependency-enhanced dual-domain parallel neural network for Ring Artifact Removal (RAR). First, based on the fact that the features of ring artifacts are different in Cartesian and Polar coordinates, the parallel architecture is adopted to construct the deep neural network so that it can extract and exploit the latent features from different domains to improve the performance of ring artifact removal. Besides, the ring artifacts are globally relevant whether in Cartesian or Polar coordinate systems, but convolutional neural networks show inherent shortcomings in modeling long-range dependency. To tackle this problem, this study introduces the novel Mamba mechanism to achieve a global receptive field without incurring high computational complexity. It enables effective capture of the long-range dependency, thereby enhancing the model performance in image restoration and artifact reduction. The experiments on the simulated data validate the effectiveness of the dual-domain parallel neural network and the Mamba mechanism, and the results on two unseen real datasets demonstrate the promising performance of the proposed RAR algorithm in eliminating ring artifacts and recovering image details.

Long scan time significantly hinders the widespread applications of three-dimensional multi-contrast cardiac magnetic resonance (3D-MC-CMR) imaging. This study aims to accelerate 3D-MC-CMR acquisition by a novel method based on score-based diffusion models with self-supervised learning. Specifically, we first establish a mapping between the undersampled k-space measurements and the MR images, utilizing a self-supervised Bayesian reconstruction network. Secondly, we develop a joint score-based diffusion model on 3D-MC-CMR images to capture their inherent distribution. The 3D-MC-CMR images are finally reconstructed using the conditioned Langenvin Markov chain Monte Carlo sampling. This approach enables accurate reconstruction without fully sampled training data. Its performance was tested on the dataset acquired by a 3D joint myocardial $ \text {T}_{{1}}$ and $ \text {T}_{{1}\rho }$ mapping sequence. The $ \text {T}_{{1}}$ and $ \text {T}_{{1}\rho }$ maps were estimated via a dictionary matching method from the reconstructed images. Experimental results show that the proposed method outperforms traditional compressed sensing and existing self-supervised deep learning MRI reconstruction methods. It also achieves high quality $ \text {T}_{{1}}$ and $ \text {T}_{{1}\rho }$ parametric maps close to the reference maps, even at a high acceleration rate of 14.

Brain lesion segmentation is crucial for neurological disease research and diagnosis. As different types of lesions exhibit distinct characteristics on different imaging modalities, segmentation methods are typically developed in a task-specific manner, where each segmentation model is tailored to a specific lesion type and modality. However, the use of task-specific models requires predetermination of the lesion type and imaging modality, which complicates their deployment in real-world scenarios. In this work, we propose a universal foundation model for brain lesion segmentation on magnetic resonance imaging (MRI), which can automatically segment different types of brain lesions given input of various MRI modalities. We develop a novel Mixture of Modality Experts (MoME) framework with multiple expert networks attending to different imaging modalities. A hierarchical gating network is proposed to combine the expert predictions and foster expertise collaboration. Moreover, to avoid the degeneration of each expert network, we introduce a curriculum learning strategy during training to preserve the specialisation of each expert. In addition to MoME, to handle the combination of multiple input modalities, we propose MoME+, which uses a soft dispatch network for input modality routing. We evaluated the proposed method on nine brain lesion datasets, encompassing five imaging modalities and eight lesion types. The results show that our model outperforms state-of-the-art universal models for brain lesion segmentation and achieves promising generalisation performance onto unseen datasets.

Accurately identifying individuals who are at high risk of lung cancer is critical to optimize lung cancer screening with low-dose CT (LDCT). We sought to compare the performance of traditional regression models and artificial intelligence (AI)-based models in predicting future lung cancer risk. A systematic review and meta-analysis were conducted with reporting according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We searched MEDLINE, Embase, Scopus, and the Cumulative Index to Nursing and Allied Health Literature databases for studies reporting the performance of AI or traditional regression models for predicting lung cancer risk. Two researchers screened articles, and a third researcher resolved conflicts. Model characteristics and predictive performance metrics were extracted. The quality of studies was assessed using the Prediction model Risk of Bias Assessment Tool. A meta-analysis assessed the discrimination performance of models, based on area under the receiver operating characteristic curve (AUC). One hundred forty studies met inclusion criteria and included 185 traditional and 64 AI-based models. Of these, 16 AI models and 65 traditional models have been externally validated. The pooled AUC of external validations of AI models was 0.82 (95% confidence interval [CI], 0.80-0.85), and the pooled AUC for traditional regression models was 0.73 (95% CI, 0.72-0.74). In a subgroup analysis, AI models that included LDCT had a pooled AUC of 0.85 (95% CI, 0.82-0.88). Overall risk of bias was high for both AI and traditional models. AI-based models, particularly those using imaging data, show promise for improving lung cancer risk prediction over traditional regression models. Future research should focus on prospective validation of AI models and direct comparisons with traditional methods in diverse populations.