Fairness concerns stemming from known and unknown biases in healthcare practices have raised questions about the trustworthiness of Artificial Intelligence (AI)-driven Clinical Decision Support Systems (CDSS). Studies have shown unforeseen performance disparities in subpopulations when applied to clinical settings different from training. Existing unfairness mitigation strategies often struggle with scalability and accessibility, while their pursuit of group-level prediction performance parity does not effectively translate into fairness at the point of care. This study introduces FairICP, a flexible and cost-effective post-implementation framework based on Inductive Conformal Prediction (ICP), to provide users with actionable knowledge of model uncertainty due to subpopulation level biases at the point of care. FairICP applies ICP to identify the model's scope of competence through group specific calibration, ensuring equitable prediction reliability by filtering predictions that fall within the trusted competence boundaries. We evaluated FairICP against four benchmarks on three medical imaging modalities: (1) Cardiac Magnetic Resonance Imaging (MRI), (2) Chest X-ray and (3) Dermatology Imaging, acquired from both private and large public datasets. Frameworks are assessed on prediction performance enhancement and unfairness mitigation capabilities. Compared to the baseline, FairICP improved prediction accuracy by 7.2% and reduced the accuracy gap between the privileged and unprivileged subpopulations by 2.2% on average across all three datasets. Our work provides a robust solution to promote trust and transparency in AI-CDSS, fostering equality and equity in healthcare for diverse patient populations. Such post-process methods are critical to enabling a robust framework for AI-CDSS implementation and monitoring for healthcare settings.

Risk stratification is a key tool in clinical decision-making, yet current approaches often fail to translate sophisticated survival analysis into actionable clinical criteria. We present a novel method for unsupervised machine learning that directly optimizes for survival heterogeneity across patient clusters through a differentiable adaptation of the multivariate logrank statistic. Unlike most existing methods that rely on proxy metrics, our approach represents novel methodology for training any neural network architecture on any data modality to identify prognostically distinct patient groups. We thoroughly evaluate the method in simulation experiments and demonstrate its utility in practice by applying it to two distinct cancer types: analyzing laboratory parameters from multiple myeloma patients and computed tomography images from non-small cell lung cancer patients, identifying prognostically distinct patient subgroups with significantly different survival outcomes in both cases. Post-hoc explainability analyses uncover clinically meaningful features determining the group assignments which align well with established risk factors and thus lend strong weight to the methods utility. This pan-cancer, model-agnostic approach represents a valuable advancement in clinical risk stratification, enabling the discovery of novel prognostic signatures across diverse data types while providing interpretable results that promise to complement treatment personalization and clinical decision-making in oncology and beyond.

Accurate segmentation of vascular structures in coronary angiography remains a core challenge in medical image analysis due to the complexity of elongated, thin, and low-contrast vessels. Classical convolutional neural networks (CNNs) often fail to preserve topological continuity, while recent Vision Transformer (ViT)-based models, although strong in global context modeling, lack precise boundary awareness. In this work, we introduce BAVT, a Boundary-Aware Vision Transformer, a ViT-based architecture enhanced with an edge-aware loss that explicitly guides the segmentation toward fine-grained vascular boundaries. Unlike hybrid transformer-CNN models, BAVT retains a minimal, scalable structure that is fully compatible with large-scale vision foundation model (VFM) pretraining. We validate our approach on the DCA-1 coronary angiography dataset, where BAVT achieves superior performance across medical image segmentation metrics outperforming both CNN and hybrid baselines. These results demonstrate the effectiveness of combining plain ViT encoders with boundary-aware supervision for clinical-grade vascular segmentation.

Generative models based on deep learning have shown significant potential in medical imaging, particularly for modality transformation and multimodal fusion in MRI-based brain imaging. This study introduces GM-LDM, a novel framework that leverages the latent diffusion model (LDM) to enhance the efficiency and precision of MRI generation tasks. GM-LDM integrates a 3D autoencoder, pre-trained on the large-scale ABCD MRI dataset, achieving statistical consistency through KL divergence loss. We employ a Vision Transformer (ViT)-based encoder-decoder as the denoising network to optimize generation quality. The framework flexibly incorporates conditional data, such as functional network connectivity (FNC) data, enabling personalized brain imaging, biomarker identification, and functional-to-structural information translation for brain diseases like schizophrenia.

Pulmonary arterial hypertension (PAH) significantly affects the pulmonary vasculature, requiring accurate estimation of mean pulmonary arterial pressure (mPAP) and pulmonary vascular resistance index (PVRi). Although cardiac catheterization is the gold standard for these measurements, it poses risks, especially in children. This pilot study explored how machine learning (ML) can predict pulmonary hemodynamics from non-invasive cardiac magnetic resonance (CMR) cine images in pediatric PAH patients. A retrospective analysis of 40 CMR studies from children with PAH using a four-fold stratified group cross-validation was conducted. The endpoints were severity profiles of mPAP and PVRi, categorised as 'low', 'high', and 'extreme'. Deep learning (DL) and traditional ML models were optimized through hyperparameter tuning. Receiver operating characteristic curves and area under the curve (AUC) were used as the primary evaluation metrics. DL models utilizing CMR cine imaging showed the best potential for predicting mPAP and PVRi severity profiles on test folds (AUC_mPAP=0.82 and AUC_PVRi=0.73). True positive rates (TPR) for predicting low, high, and extreme mPAP were 5/10, 11/16, and 11/14, respectively. TPR for predicting low, high, and extreme PVRi were 5/13, 14/15, and 7/12, respectively. Optimal DL models only used spatial patterns from consecutive CMR cine frames to maximize prediction performance. This exploratory pilot study demonstrates the potential of DL leveraging CMR imaging for non-invasive prediction of mPAP and PVRi in pediatric PAH. While preliminary, these findings may lay the groundwork for future advancements in CMR imaging in pediatric PAH, offering a pathway to safer disease monitoring and reduced reliance on invasive cardiac catheterization.

To evaluate a deep learning sequence-adaptive liver multiparametric MRI (mpMRI) assessment with validation in different populations using total and segmental T1 and T2 relaxation time maps. A neural network was trained to label liver segmental parenchyma and its vessels on noncontrast T1-weighted gradient-echo Dixon in-phase acquisitions on 200 liver mpMRI examinations. Then, 120 unseen liver mpMRI examinations of patients with primary sclerosing cholangitis or healthy controls were assessed by coregistering the labels to noncontrast and contrast-enhanced T1 and T2 relaxation time maps for optimization and internal testing. The algorithm was externally tested in a segmental and total liver analysis of previously unseen 65 patients with biopsy-proven liver fibrosis and 25 healthy volunteers. Measured relaxation times were compared to manual measurements using intraclass correlation coefficient (ICC) and Wilcoxon test. Comparison of manual and deep learning-generated segmental areas on different T1 and T2 maps was excellent for segmental (ICC = 0.95 ± 0.1; p < 0.001) and total liver assessment (0.97 ± 0.02, p < 0.001). The resulting median of the differences between automated and manual measurements among all testing populations and liver segments was 1.8 ms for noncontrast T1 (median 835 versus 842 ms), 2.0 ms for contrast-enhanced T1 (median 518 versus 519 ms), and 0.3 ms for T2 (median 37 versus 37 ms). Automated quantification of liver mpMRI is highly effective across different patient populations, offering excellent reliability for total and segmental T1 and T2 maps. Its scalable, sequence-adaptive design could foster comprehensive clinical decision-making. The proposed automated, sequence-adaptive algorithm for total and segmental analysis of liver mpMRI may be co-registered to any combination of parametric sequences, enabling comprehensive quantitative analysis of liver mpMRI without sequence-specific training. A deep learning-based algorithm automatically quantified segmental T1 and T2 relaxation times in liver mpMRI. The two-step approach of segmentation and co-registration allowed to assess arbitrary sequences. The algorithm demonstrated high reliability with manual reader quantification. No additional sequence-specific training is required to assess other parametric sequences. The DL algorithm has the potential to enhance individual liver phenotyping.

Artificial intelligence (AI) systems for age-related macular degeneration (AMD) diagnosis abound but are not yet widely implemented. AI implementation is complex, requiring the involvement of multiple, diverse stakeholders including technology developers, clinicians, patients, health networks, public hospitals, private providers and payers. There is a pressing need to investigate how AI might be adopted to improve patient outcomes. The purpose of this first study of its kind was to use the AI translation extended version of the non-adoption, abandonment, scale-up, spread and sustainability of healthcare technologies framework to explore stakeholder experiences, attitudes, enablers, barriers and possible futures of digital diagnosis using AI for AMD and eyecare in Australia. Semi-structured, online interviews were conducted with 37 stakeholders (12 clinicians, 10 healthcare leaders, 8 patients and 7 developers) from September 2022 to March 2023. The interviews were audio-recorded, transcribed and analysed using directed and summative content analysis. Technological features influencing implementation were most frequently discussed, followed by the context or wider system, value proposition, adopters, organisations, the condition and finally embedding the adaptation. Patients preferred to focus on the condition, while healthcare leaders elaborated on organisation factors. Overall, stakeholders supported a portable, device-independent clinical decision support tool that could be integrated with existing diagnostic equipment and patient management systems. Opportunities for AI to drive new models of healthcare, patient education and outreach, and the importance of maintaining equity across population groups were consistently emphasised. This is the first investigation to report numerous, interacting perspectives on the adoption of digital diagnosis for AMD in Australia, incorporating an intentionally diverse stakeholder group and the patient voice. It provides a series of practical considerations for the implementation of AI and digital diagnosis into existing care for people with AMD.

Accurate segmentation of pancreatic ductal adenocarcinoma (PDAC) and surrounding anatomical structures is critical for diagnosis, treatment planning, and outcome assessment. This study proposes a deep learning-based framework to automate multi-class segmentation in CT images, comparing the performance of four state-of-the-art architectures. This retrospective multi-center study included 3265 patients from six institutions. Four deep learning models-UNet, nnU-Net, UNETR, and Swin-UNet-were trained using five-fold cross-validation on data from five centers and tested independently on a sixth center (n = 569). Preprocessing included intensity normalization, voxel resampling, and standardized annotation for six structures: PDAC lesion, pancreas, veins, arteries, pancreatic duct, and common bile duct. Evaluation metrics included Dice Similarity Coefficient (DSC), Intersection over Union (IoU), directed Hausdorff Distance (dHD), Average Symmetric Surface Distance (ASSD), and Volume Overlap Error (VOE). Statistical comparisons were made using Wilcoxon signed-rank tests with Bonferroni correction. Swin-UNet outperformed all models with a mean validation DSC of 92.4% and test DSC of 90.8%, showing minimal overfitting. It also achieved the lowest dHD (4.3 mm), ASSD (1.2 mm), and VOE (6.0%) in cross-validation. Per-class DSCs for Swin-UNet were consistently higher across all anatomical targets, including challenging structures like the pancreatic duct (91.0%) and bile duct (91.8%). Statistical analysis confirmed the superiority of Swin-UNet (p < 0.001). All models showed generalization capability, but Swin-UNet provided the most accurate and robust segmentation across datasets. Transformer-based architectures, particularly Swin-UNet, enable precise and generalizable multi-class segmentation of PDAC and surrounding anatomy. This framework has potential for clinical integration in PDAC diagnosis, staging, and therapy planning.

Segmentation of target volumes and organs at risk on computed tomography (CT) images constitutes an important step in the radiotherapy workflow. Artificial intelligence-based methods have significantly improved organ segmentation in medical images. Automatic segmentations are frequently evaluated using geometric metrics. Before a clinical implementation in the radiotherapy workflow, automatic segmentations must also be evaluated by clinicians. The aim of this study was to investigate the correlation between geometric metrics used for segmentation evaluation and the assessment performed by clinicians. In this study, we used the U-Net model to segment the liver in CT images from a publicly available dataset. The model's performance was evaluated using two geometric metrics: the Dice similarity coefficient and the Hausdorff distance. Additionally, a qualitative evaluation was performed by clinicians who reviewed the automatic segmentations to rate their clinical acceptability for use in the radiotherapy workflow. The correlation between the geometric metrics and the clinicians' evaluations was studied. The results showed that while the Dice coefficient and Hausdorff distance are reliable indicators of segmentation accuracy, they do not always align with clinician segmentation. In some cases, segmentations with high Dice scores still required clinician corrections before clinical use in the radiotherapy workflow. This study highlights the need for more comprehensive evaluation metrics beyond geometric measures to assess the clinical acceptability of artificial intelligence-based segmentation. Although the deep learning model provided promising segmentation results, the present study shows that standardized validation methodologies are crucial for ensuring the clinical viability of automatic segmentation systems.

Early identification of unresectable hepatocellular carcinoma (HCC) patients who may benefit from immune checkpoint inhibitors (ICIs) is crucial for optimizing outcomes. Here, we developed a multimodal fusion (MMF) system integrating CT-derived deep learning features and clinical data to predict overall survival (OS) and progression-free survival (PFS). Using retrospective multicenter data (n = 859), the MMF combining an ensemble deep learning (Ensemble-DL) model with clinical variables achieved strong external validation performance (C-index: OS = 0.74, PFS = 0.69), outperforming radiomics (29.8% OS improvement), mRECIST (27.6% OS improvement), clinical benchmarks (C-index: OS = 0.67, p = 0.0011; PFS = 0.65, p = 0.033), and Ensemble-DL (C-index: OS = 0.69, p = 0.0028; PFS = 0.66, p = 0.044). The MMF system effectively stratified patients across clinical subgroups and demonstrated interpretability through activation maps and radiomic correlations. Differential gene expression analysis revealed enrichment of the PI3K/Akt pathway in patients identified by the MMF system. The MMF system provides an interpretable, clinically applicable approach to guide personalized ICI treatment in unresectable HCC.