Artificial Intelligence (AI) holds significant promise for improving prognosis prediction in medical imaging, yet its effective application remains challenging. In this work, we introduce a structured benchmark explicitly designed to evaluate and compare the transferability of Convolutional Neural Networks and Foundation Models in predicting clinical outcomes in COVID-19 patients, leveraging diverse publicly available Chest X-ray datasets. Our experimental methodology extensively explores a wide set of fine-tuning strategies, encompassing traditional approaches such as Full Fine-Tuning and Linear Probing, as well as advanced Parameter-Efficient Fine-Tuning methods including Low-Rank Adaptation, BitFit, VeRA, and IA3. The evaluations were conducted across multiple learning paradigms, including both extensive full-data scenarios and more clinically realistic Few-Shot Learning settings, which are critical for modeling rare disease outcomes and rapidly emerging health threats. By implementing a large-scale comparative analysis involving a diverse selection of pretrained models, including general-purpose architectures pretrained on large-scale datasets such as CLIP and DINOv2, to biomedical-specific models like MedCLIP, BioMedCLIP, and PubMedCLIP, we rigorously assess each model's capacity to effectively adapt and generalize to prognosis tasks, particularly under conditions of severe data scarcity and pronounced class imbalance. The benchmark was designed to capture critical conditions common in prognosis tasks, including variations in dataset size and class distribution, providing detailed insights into the strengths and limitations of each fine-tuning strategy. This extensive and structured evaluation aims to inform the practical deployment and adoption of robust, efficient, and generalizable AI-driven solutions in real-world clinical prognosis prediction workflows.

B-line artifacts in lung ultrasound, pivotal for diagnosing pulmonary conditions, warrant automated recognition to enhance diagnostic accuracy. In this paper, a lung ultrasound B-line vertical artifact identification method based on radio frequency (RF) signal was proposed. B-line regions were distinguished from non-B-line regions by inputting multiple characteristic parameters into nonlinear support vector machine (SVM). Six characteristic parameters were evaluated, including permutation entropy, information entropy, kurtosis, skewness, Nakagami shape factor, and approximate entropy. Following the evaluation that demonstrated the performance differences in parameter recognition, Principal Component Analysis (PCA) was utilized to reduce the dimensionality to a four-dimensional feature set for input into a nonlinear Support Vector Machine (SVM) for classification purposes. Four types of experiments were conducted: a sponge with dripping water model, gelatin phantoms containing either glass beads or gelatin droplets, and in vivo experiments. By employing precise feature selection and analyzing scan lines rather than full images, this approach significantly reduced the dependency on large image datasets without compromising discriminative accuracy. The method exhibited performance comparable to contemporary image-based deep learning approaches, which, while highly effective, typically necessitate extensive data for training and require expert annotation of large datasets to establish ground truth. Owing to the optimized architecture of our model, efficient sample recognition was achieved, with the capability to process between 27,000 and 33,000 scan lines per second (resulting in a frame rate exceeding 100 FPS, with 256 scan lines per frame), thus supporting real-time analysis. The results demonstrate that the accuracy of the method to classify a scan line as belonging to a B-line region was up to 88%, with sensitivity reaching up to 90%, specificity up to 87%, and an F1-score up to 89%. This approach effectively reflects the performance of scan line classification pertinent to B-line identification. Our approach reduces the reliance on large annotated datasets, thereby streamlining the preprocessing phase.

Cone-beam computed tomography (CBCT) using only a few X-ray projection views enables faster scans with lower radiation dose, but the resulting severe under-sampling causes strong artifacts and poor spatial coverage. We address these challenges in a unified framework. First, we replace conventional UNet/ResNet encoders with TransUNet, a hybrid CNN-Transformer model. Convolutional layers capture local details, while self-attention layers enhance global context. We adapt TransUNet to CBCT by combining multi-scale features, querying view-specific features per 3D point, and adding a lightweight attenuation-prediction head. This yields Trans-CBCT, which surpasses prior baselines by 1.17 dB PSNR and 0.0163 SSIM on the LUNA16 dataset with six views. Second, we introduce a neighbor-aware Point Transformer to enforce volumetric coherence. This module uses 3D positional encoding and attention over k-nearest neighbors to improve spatial consistency. The resulting model, Trans$^2$-CBCT, provides an additional gain of 0.63 dB PSNR and 0.0117 SSIM. Experiments on LUNA16 and ToothFairy show consistent gains from six to ten views, validating the effectiveness of combining CNN-Transformer features with point-based geometry reasoning for sparse-view CBCT reconstruction.

The COVID-19 pandemic plays a significant roles in the global health, highlighting the imperative for effective management of post-recovery symptoms. Within this context, Ground Glass Opacity (GGO) in lung computed tomography CT scans emerges as a critical indicator for early intervention. Recently, most researchers have investigated initially a challenge to refine techniques for GGO segmentation. These approaches aim to scrutinize and juxtapose cutting-edge methods for analyzing lung CT images of patients recuperating from COVID-19. While many methods in this challenge utilize the nnU-Net architecture, its general approach has not concerned completely GGO areas such as marking infected areas, ground-glass opacity, irregular shapes and fuzzy boundaries. This research has investigated a specialized machine learning algorithm, advancing the nn-UNet framework to accurately segment GGO in lung CT scans of post-COVID-19 patients. We propose a novel approach for two-stage image segmentation methods based on nnU-Net 2D and 3D models including lung and shadow image segmentation, incorporating the attention mechanism. The combination models enhance automatic segmentation and models' accuracy when using different error function in the training process. Experimental results show that the proposed model's outcomes DSC score ranks fifth among the compared results. The proposed method has also the second-highest sensitivity value among the methods, which shows that this method has a higher true segmentation rate than most of the other methods. The proposed method has achieved a Hausdorff95 of 54.566, Surface dice of 0.7193, Sensitivity of 0.7528, and Specificity of 0.7749. As compared with the state-of-the-art methods, the proposed model in experimental results is improved much better than the current methods in term of segmentation of infected areas. The proposed model has been deployed in the case study of real-world problems with the combination of 2D and 3D models. It is demonstrated the capacity to comprehensively detect lung lesions correctly. Additionally, the boundary loss function has assisted in achieving more precise segmentation for low-resolution images. Initially segmenting lung area has reduced the volume of images requiring processing, while diminishing for training process.

Considerable challenges exist in managing lung cancer and mesothelioma, including diagnostic complexity, treatment stratification, early detection and imaging quantification. Variable incidence in mesothelioma also makes equitable provision of high-quality care difficult. In this context, artificial intelligence (AI) offers a range of assistive/automated functions that can potentially enhance clinical decision-making, while reducing inequality and pathway delay. In this state-of-the-art narrative review, we synthesise evidence on this topic, focusing particularly on tools that ingest routine pathology and radiology images. We summarise the strengths and weaknesses of AI applied to common multidisciplinary team (MDT) functions, including histological diagnosis, therapeutic response prediction, radiological detection and quantification, and survival estimation. We also review emerging methods capable of generating novel biological insights and current barriers to implementation, including access to high-quality training data and suitable regulatory and technical infrastructure. Neural networks trained on pathology images have proven utility in histological classification, prognostication, response prediction and survival. Self-supervised models can also generate new insights into biological features responsible for adverse outcomes. Radiology applications include lung nodule tools, which offer critical pathway support for imminent lung cancer screening and urgent referrals. Tumour segmentation AI offers particular advantages in mesothelioma, where response assessment and volumetric staging are difficult using human readers due to tumour size and morphological complexity. AI is also critical for radiogenomics, permitting effective integration of molecular and radiomic features for discovery of non-invasive markers for molecular subtyping and enhanced stratification. AI solutions offer considerable potential benefits across the MDT, particularly in repetitive or time-consuming tasks based on pathology and radiology images. Effective leveraging of this technology is critical for lung cancer screening and efficient delivery of increasingly complex diagnostic and predictive MDT functions. Future AI research should involve transparent and interpretable outputs that assist in explaining the basis of AI-supported decision making.

High-resolution computed tomography (HRCT) is helpful for diagnosing interstitial lung diseases (ILD), but it largely depends on the experience of physicians. Herein, our study aims to develop a deep-learning-based classification model to differentiate the three common types of ILD, so as to provide a reference to help physicians make the diagnosis and improve the accuracy of ILD diagnosis. Patients were selected from four tertiary Grade A hospitals in Kunming based on inclusion and exclusion criteria. HRCT scans of 130 patients were included. The imaging manifestations were usual interstitial pneumonia (UIP), non-specific interstitial pneumonia (NSIP), and organizing pneumonia (OP). Additionally, 50 chest HRCT cases without imaging abnormalities during the same period were selected.Construct a data set. Conduct the training, validation, and testing of the Parallel Multi-scale Feature Fusion Network (PMFF-Net) deep learning model. Utilize Python software to generate data and charts pertaining to model performance. Assess the model's accuracy, precision, recall, and F1-score, and juxtapose its diagnostic efficacy against that of physicians across various hospital levels, with differing levels of seniority, and from various departments. The PMFF -Net deep learning model is capable of classifying imaging types such as UIP, NSIP, and OP, as well as normal imaging. In a mere 105 s, it makes the diagnosis for 18 HRCT images with a diagnostic accuracy of 92.84 %, precision of 91.88 %, recall of 91.95 %, and an F1 score of 0.9171. The diagnostic accuracy of senior radiologists (83.33 %) and pulmonologists (77.77 %) from tertiary hospitals is higher than that of internists from secondary hospitals (33.33 %). Meanwhile, the diagnostic accuracy of middle-aged radiologists (61.11 %) and pulmonologists (66.66 %) are higher than junior radiologists (38.88 %) and pulmonologists (44.44 %) in tertiary hospitals, whereas junior and middle-aged internists at secondary hospitals were unable to complete the tests. This study found that the PMFF-Net model can effectively classify UIP, NSIP, OP imaging types, and normal imaging, which can help doctors of different hospital levels and departments make clinical decisions quickly and effectively.

During radiography, irregularities occur, leading to decrease in the diagnostic value of the images obtained. The purpose of this work was to develop a system for automated quality assurance of patient positioning in chest radiographs, with detection of suboptimal contrast, brightness, and metadata errors. The quality assurance system was trained and tested using more than 69,000 X-rays of the chest and other anatomical areas from the Unified Radiological Information Service (URIS) and several open datasets. Our dataset included studies regardless of a patient's gender and race, while the sole exclusion criterion being age below 18 years. A training dataset of radiographs labeled by expert radiologists was used to train an ensemble of modified deep convolutional neural networks architectures ResNet152V2 and VGG19 to identify various quality deficiencies. Model performance was accessed using area under the receiver operating characteristic curve (ROC-AUC), precision, recall, F1-score, and accuracy metrics. Seven neural network models were trained to classify radiographs by the following quality deficiencies: failure to capture the target anatomic region, chest rotation, suboptimal brightness, incorrect anatomical area, projection errors, and improper photometric interpretation. All metrics for each model exceed 95%, indicating high predictive value. All models were combined into a unified system for evaluating radiograph quality. The processing time per image is approximately 3 s. The system supports multiple use cases: integration into an automated radiographic workstations, external quality assurance system for radiology departments, acquisition quality audits for municipal health systems, and routing of studies to diagnostic AI models.

To develop and validate a CT image-based multiple time-series deep learning model for the longitudinal prediction of benign and malignant pulmonary ground-glass nodules (GGNs). A total of 486 GGNs from an equal number of patients were included in this research, which took place at two medical centers. Each nodule underwent surgical removal and was confirmed pathologically. The patients were randomly assigned to a training set, validation set, and test set, following a distribution ratio of 7:2:1. We established a transformer-based deep learning framework that leverages multi-temporal CT images for the longitudinal prediction of GGNs, focusing on distinguishing between benign and malignant types. Additionally, we utilized 13 different machine learning algorithms to formulate clinical models, delta-radiomics models, and combined models that merge deep learning with CT semantic features. The predictive capabilities of the models were assessed using the receiver operating characteristic (ROC) curve and the area under the curve (AUC). The multiple time-series deep learning model based on CT images surpassed both the clinical model and the delta-radiomics model, showcasing strong predictive capabilities for GGNs across the training, validation, and test sets, with AUCs of 0.911 (95% CI, 0.879-0.939), 0.809 (95% CI,0.715-0.908), and 0.817 (95% CI,0.680-0.937), respectively. Furthermore, the models that integrated deep learning with CT semantic features achieved the highest performance, resulting in AUCs of 0.960 (95% CI, 0.912-0.977), 0.878 (95% CI,0.801-0.942), and 0.890(95% CI, 0.790-0.968). The multiple time-series deep learning model utilizing CT images was effective in predicting benign and malignant GGNs.

Lung cancer remains the leading cause of cancer-related mortality worldwide, due to lacking effective early-stage screening approaches. Imaging, such as low-dose CT, poses radiation risk, and biopsies can induce some complications. Additionally, traditional serum tumor markers lack diagnostic specificity. This highlights the urgent need for precise and non-invasive early detection techniques. This systematic review aims to evaluate the limitations of conventional screening methods (imaging/biopsy/tumor markers), seek breakthroughs in liquid biopsy for early lung cancer detection, and assess the potential value of Artificial Intelligence (AI), thereby providing evidence-based insights for establishing an optimal screening framework. We systematically searched the PubMed database for the literature published up to May 2025. Key words include "Artificial Intelligence", "Early Lung cancer screening", "Imaging examination", "Innovative technologies", "Liquid biopsy", and "Puncture biopsy". Our inclusion criteria focused on studies about traditional and innovative screening methods, with an emphasis on original research concerning diagnostic performance or high-quality reviews. This approach helps identify critical studies in early lung cancer screening. Novel liquid biopsy techniques are non-invasive and have superior diagnostic efficacy. AI-assisted diagnostics further enhance accuracy. We propose three development directions: establishing risk-based liquid biopsy screening protocols, developing a stepwise "imaging-AI-liquid biopsy" diagnostic workflow, and creating standardized biomarker panel testing solutions. Integrating traditional methodologies, novel liquid biopsies, and AI to establish a comprehensive early lung cancer screening model is important. These innovative strategies aim to significantly increase early detection rates, substantially enhancing lung cancer control. This review provides both theoretical guidance for clinical practice and future research.

BackgroundCT-based radio-biomarkers could provide non-invasive insights into tumour biology to risk-stratify patients. One of the limitations is laborious manual segmentation of regions-of-interest (ROI). We present a deep learning auto-segmentation pipeline for radiomic analysis. Patients and Methods153 patients with resected stage 2A-3B non-small cell lung cancer (NSCLCs) had tumours segmented using nnU-Net with review by two clinicians. The nnU-Net was pretrained with anatomical priors in non-cancerous lungs and finetuned on NSCLCs. Three ROIs were segmented: intra-tumoural, peri-tumoural, and whole lung. 1967 features were extracted using PyRadiomics. Feature reproducibility was tested using segmentation perturbations. Features were selected using minimum-redundancy-maximum-relevance with Random Forest-recursive feature elimination nested in 500 bootstraps. ResultsAuto-segmentation time was [~]36 seconds/series. Mean volumetric and surface Dice-Sorensen coefficient (DSC) scores were 0.84 ({+/-}0.28), and 0.79 ({+/-}0.34) respectively. DSC were significantly correlated with tumour shape (sphericity, diameter) and location (worse with chest wall adherence), but not batch effects (e.g. contrast, reconstruction kernel). 6.5% cases had missed segmentations; 6.5% required major changes. Pre-training on anatomical priors resulted in better segmentations compared to training on tumour-labels alone (p<0.001) and tumour with anatomical labels (p<0.001). Most radiomic features were not reproducible following perturbations and resampling. Adding radiomic features, however, did not significantly improve the clinical model in predicting 2-year disease-free survival: AUCs 0.67 (95%CI 0.59-0.75) vs 0.63 (95%CI 0.54-0.71) respectively (p=0.28). ConclusionOur study demonstrates that integrating auto-segmentation into radio-biomarker discovery is feasible with high efficiency and accuracy. Whilst radiomic analysis show limited reproducibility, our auto-segmentation may allow more robust radio-biomarker analysis using deep learning features.