3D medical image segmentation is vital for clinical diagnosis and treatment but is challenged by high-dimensional data and complex spatial dependencies. Traditional single-modality networks, such as CNNs and Transformers, are often limited by computational inefficiency and constrained contextual modeling in 3D settings. We introduce a novel multimodal framework that leverages Mamba and Kolmogorov-Arnold Networks (KAN) as an efficient backbone for long-sequence modeling. Our approach features three key innovations: First, an EGSC (Enhanced Gated Spatial Convolution) module captures spatial information when unfolding 3D images into 1D sequences. Second, we extend Group-Rational KAN (GR-KAN), a Kolmogorov-Arnold Networks variant with rational basis functions, into 3D-Group-Rational KAN (3D-GR-KAN) for 3D medical imaging - its first application in this domain - enabling superior feature representation tailored to volumetric data. Third, a dual-branch text-driven strategy leverages CLIP's text embeddings: one branch swaps one-hot labels for semantic vectors to preserve inter-organ semantic relationships, while the other aligns images with detailed organ descriptions to enhance semantic alignment. Experiments on the Medical Segmentation Decathlon (MSD) and KiTS23 datasets show our method achieving state-of-the-art performance, surpassing existing approaches in accuracy and efficiency. This work highlights the power of combining advanced sequence modeling, extended network architectures, and vision-language synergy to push forward 3D medical image segmentation, delivering a scalable solution for clinical use. The source code is openly available at https://github.com/yhy-whu/TK-Mamba.

Segmentation of 3D medical images, and brain segmentation in particular, is an important topic in neuroimaging and in radiotherapy. Overcoming the current, time consuming, practise of manual delineation of brain tumours and providing an accurate, explainable, and replicable method of segmentation of the tumour area and related tissues is therefore an open research challenge. In this paper, we first propose a novel symbolic approach to brain segmentation and delineation of brain lesions based on spatial model checking. This method has its foundations in the theory of closure spaces, a generalisation of topological spaces, and spatial logics. At its core is a high-level declarative logic language for image analysis, ImgQL, and an efficient spatial model checker, VoxLogicA, exploiting state-of-the-art image analysis libraries in its model checking algorithm. We then illustrate how this technique can be combined with Machine Learning techniques leading to a hybrid AI approach that provides accurate and explainable segmentation results. We show the results of the application of the symbolic approach on several public datasets with 3D magnetic resonance (MR) images. Three datasets are provided by the 2017, 2019 and 2020 international MICCAI BraTS Challenges with 210, 259 and 293 MR images, respectively, and the fourth is the BrainWeb dataset with 20 (synthetic) 3D patient images of the normal brain. We then apply the hybrid AI method to the BraTS 2020 training set. Our segmentation results are shown to be in line with the state-of-the-art with respect to other recent approaches, both from the accuracy point of view as well as from the view of computational efficiency, but with the advantage of them being explainable.

Alzheimer's disease (AD) is a neurodegenerative disorder characterized by a progressive cognitive decline, necessitating early diagnosis for effective treatment. This study presents the Relational Bi-level Aggregation Graph Convolutional Network with Dynamic Graph Learning and Puzzle Optimization for Alzheimer's Classification (RBAGCN-DGL-PO-AC), using denoised T1-weighted Magnetic Resonance Images (MRIs) collected from Alzheimer's Disease Neuroimaging Initiative (ADNI) repository. Addressing the impact of noise in medical imaging, the method employs advanced denoising techniques includes: the Modified Spline-Kernelled Chirplet Transform (MSKCT), Jump Gain Integral Recurrent Neural Network (JGIRNN), and Newton Time Extracting Wavelet Transform (NTEWT), to enhance the image quality. Key brain regions, crucial for classification such as hippocampal, lateral ventricle and posterior cingulate cortex are segmented using Attention Guided Generalized Intuitionistic Fuzzy C-Means Clustering (AG-GIFCMC). Feature extraction and classification using segmented outputs are performed with RBAGCN-DGL and puzzle optimization, categorize input images into Healthy Controls (HC), Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI), and Alzheimer's Disease (AD). To assess the effectiveness of the proposed method, we systematically examined the structural modifications to the RBAGCN-DGL-PO-AC model through extensive ablation studies. Experimental findings highlight that RBAGCN-DGL-PO-AC state-of-the art performance, with 99.25 % accuracy, outperforming existing methods including MSFFGCN_ADC, CNN_CAD_DBMRI, and FCNN_ADC, while reducing training time by 28.5 % and increasing inference speed by 32.7 %. Hence, the RBAGCN-DGL-PO-AC method enhances AD classification by integrating denoising, segmentation, and dynamic graph-based feature extraction, achieving superior accuracy and making it a valuable tool for clinical applications, ultimately improving patient outcomes and disease management.

AI-based MRI reconstruction techniques improve efficiency by reducing acquisition times whilst maintaining or improving image quality. Recent recommendations from professional bodies suggest centres should perform quality assessments on AI tools. However, monitoring long-term performance presents challenges, due to model drift or system updates. Radiologist-based assessments are resource-intensive and may be subjective, highlighting the need for efficient quality control (QC) measures. This study explores using image quality metrics (IQMs) to assess AI-based reconstructions. 58 patients undergoing standard-of-care rectal MRI were imaged using AI-based and conventional T2-weighted sequences. Paired and unpaired IQMs were calculated. Sensitivity of IQMs to detect retrospective perturbations in AI-based reconstructions was assessed using control charts, and statistical comparisons between the four MR systems in the evaluation were performed. Two radiologists evaluated the image quality of the perturbed images, giving an indication of their clinical relevance. Paired IQMs demonstrated sensitivity to changes in AI-reconstruction settings, identifying deviations outside ± 2 standard deviations of the reference dataset. Unpaired metrics showed less sensitivity. Paired IQMs showed no difference in performance between 1.5 T and 3 T systems (p > 0.99), whilst minor but significant (p < 0.0379) differences were noted for unpaired IQMs. IQMs are effective for QC of AI-based MR reconstructions, offering resource-efficient alternatives to repeated radiologist evaluations. Future work should expand this to other imaging applications and assess additional measures.

Atrial fibrillation (AF) is one of the primary etiologies for ischemic stroke, and it is of paramount importance to delineate the risk phenotypes among elderly AF patients and to investigate more efficacious models for predicting stroke risk. This single-center prospective cohort study collected clinical data and cardiac computed tomography angiography (CTA) images from elderly AF patients. The clinical phenotypes and left atrial appendage (LAA) radiomic phenotypes of elderly AF patients were identified through K-means clustering. The independent correlations between these phenotypes and stroke risk were subsequently analyzed. Machine learning algorithms-Logistic Regression, Naive Bayes, Support Vector Machine (SVM), Random Forest, and Extreme Gradient Boosting-were selected to develop a predictive model for stroke risk in this patient cohort. The model was assessed using the Area Under the Receiver Operating Characteristic Curve, Hosmer-Lemeshow tests, and Decision Curve Analysis. A total of 419 elderly AF patients (≥ 65 years old) were included. K-means clustering identified three clinical phenotypes: Group A (cardiac enlargement/dysfunction), Group B (normal phenotype), and Group C (metabolic/coagulation abnormalities). Stroke incidence was highest in Group A (19.3%) and Group C (14.5%) versus Group B (3.3%). Similarly, LAA radiomic phenotypes revealed elevated stroke risk in patients with enlarged LAA structure (Group B: 20.0%) and complex LAA morphology (Group C: 14.0%) compared to normal LAA (Group A: 2.9%). Among the five machine learning models, the SVM model achieved superior prediction performance (AUROC: 0.858 [95% CI: 0.830-0.887]). The stroke-risk prediction model for elderly AF patients constructed based on the SVM algorithm has strong predictive efficacy.

To develop and validate an interpretable machine learning model based on clinicoradiological features and radiomic features based on magnetic resonance imaging (MRI) to predict the failure of conservative treatment in lateral epicondylitis (LE). This retrospective study included 420 patients with LE from three hospitals, divided into a training cohort (n = 245), an internal validation cohort (n = 115), and an external validation cohort (n = 60). Patients were categorized into conservative treatment failure (n = 133) and conservative treatment success (n = 287) groups based on the outcome of conservative treatment. We developed two predictive models: one utilizing clinicoradiological features, and another integrating clinicoradiological and radiomic features. Seven machine learning algorithms were evaluated to determine the optimal model for predicting the failure of conservative treatment. Model performance was assessed using ROC, and model interpretability was examined using SHapley Additive exPlanations (SHAP). The LightGBM algorithm was selected as the optimal model because of its superior performance. The combined model demonstrated enhanced predictive accuracy with an area under the ROC curve (AUC) of 0.96 (95% CI: 0.91, 0.99) in the external validation cohort. SHAP analysis identified the radiological feature "CET coronal tear size" and the radiomic feature "AX_log-sigma-1-0-mm-3D_glszm_SmallAreaEmphasis" as key predictors of conservative treatment failure. We developed and validated an interpretable LightGBM machine learning model that integrates clinicoradiological and radiomic features to predict the failure of conservative treatment in LE. The model demonstrates high predictive accuracy and offers valuable insights into key prognostic factors.

Accurate differentiation of the histologic invasiveness of early-stage lung adenocarcinoma is crucial for determining surgical strategies. This study aimed to investigate the potential of [⁶⁸Ga]Ga-FAPI-46 PET/CT in assessing the invasiveness of early lung adenocarcinoma presenting as ground-glass nodules (GGNs) and identifying imaging features with strong predictive potential. This prospective study (NCT04588064) was conducted between July 2020 and July 2022, focusing on GGNs that were confirmed postoperatively to be either invasive adenocarcinoma (IAC), minimally invasive adenocarcinoma (MIA), or precursor glandular lesions (PGL). A total of 45 patients with 53 pulmonary GGNs were included in the study: 19 patients with GGNs associated with PGL-MIA and 34 with IAC. Lung nodules were segmented using the Segment Anything Model in Medical Images (MedSAM) and the PET Tumor Segmentation Extension. Clinical characteristics, along with conventional and high-throughput radiomics features from High-resolution CT (HRCT) and PET scans, were analysed. The predictive performance of these features in differentiating between PGL or MIA (PGL-MIA) and IAC was assessed using 5-fold cross-validation across six machine learning algorithms. Model validation was performed on an independent external test set (n = 11). The Chi-squared, Fisher's exact, and DeLong tests were employed to compare the performance of the models. The maximum standardised uptake value (SUVmax) derived from [⁶⁸Ga]Ga-FAPI-46 PET was identified as an independent predictor of IAC. A cut-off value of 1.82 yielded a sensitivity of 94% (32/34), specificity of 84% (16/19), and an overall accuracy of 91% (48/53) in the training set, while achieving 100% (12/12) accuracy in the external test set. Radiomics-based classification further improved diagnostic performance, achieving a sensitivity of 97% (33/34), specificity of 89% (17/19), accuracy of 94% (50/53), and an area under the receiver operating characteristic curve (AUC) of 0.97 [95% CI: 0.93-1.00]. Compared with the CT-based radiomics model and the PET-based model, the combined PET/CT radiomics model did not show significant improvement in predictive performance. The key predictive feature was [⁶⁸Ga]Ga-FAPI-46 PET log-sigma-7-mm-3D_firstorder_RootMeanSquared. The SUVmax derived from [⁶⁸Ga]Ga-FAPI-46 PET/CT can effectively differentiate the invasiveness of early-stage lung adenocarcinoma manifesting as GGNs. Integrating high-throughput features from [⁶⁸Ga]Ga-FAPI-46 PET/CT images can considerably enhance classification accuracy. NCT04588064; URL: https://clinicaltrials.gov/study/NCT04588064 .

Purpose: To evaluate the impact of harmonization and multi-region CT image feature integration on survival prediction in non-small cell lung cancer (NSCLC) patients, using handcrafted radiomics, pretrained foundation model (FM) features, and clinical data from a multicenter dataset. Methods: We analyzed CT scans and clinical data from 876 NSCLC patients (604 training, 272 test) across five centers. Features were extracted from the whole lung, tumor, mediastinal nodes, coronary arteries, and coronary artery calcium (CAC). Handcrafted radiomics and FM deep features were harmonized using ComBat, reconstruction kernel normalization (RKN), and RKN+ComBat. Regularized Cox models predicted overall survival; performance was assessed using the concordance index (C-index), 5-year time-dependent area under the curve (t-AUC), and hazard ratio (HR). SHapley Additive exPlanations (SHAP) values explained feature contributions. A consensus model used agreement across top region of interest (ROI) models to stratify patient risk. Results: TNM staging showed prognostic utility (C-index = 0.67; HR = 2.70; t-AUC = 0.85). The clinical + tumor radiomics model with ComBat achieved a C-index of 0.7552 and t-AUC of 0.8820. FM features (50-voxel cubes) combined with clinical data yielded the highest performance (C-index = 0.7616; t-AUC = 0.8866). An ensemble of all ROIs and FM features reached a C-index of 0.7142 and t-AUC of 0.7885. The consensus model, covering 78% of valid test cases, achieved a t-AUC of 0.92, sensitivity of 97.6%, and specificity of 66.7%. Conclusion: Harmonization and multi-region feature integration improve survival prediction in multicenter NSCLC data. Combining interpretable radiomics, FM features, and consensus modeling enables robust risk stratification across imaging centers.

We present a fully automated, anatomically guided deep learning pipeline for prostate cancer (PCa) risk stratification using routine MRI. The pipeline integrates three key components: an nnU-Net module for segmenting the prostate gland and its zones on axial T2-weighted MRI; a classification module based on the UMedPT Swin Transformer foundation model, fine-tuned on 3D patches with optional anatomical priors and clinical data; and a VAE-GAN framework for generating counterfactual heatmaps that localize decision-driving image regions. The system was developed using 1,500 PI-CAI cases for segmentation and 617 biparametric MRIs with metadata from the CHAIMELEON challenge for classification (split into 70% training, 10% validation, and 20% testing). Segmentation achieved mean Dice scores of 0.95 (gland), 0.94 (peripheral zone), and 0.92 (transition zone). Incorporating gland priors improved AUC from 0.69 to 0.72, with a three-scale ensemble achieving top performance (AUC = 0.79, composite score = 0.76), outperforming the 2024 CHAIMELEON challenge winners. Counterfactual heatmaps reliably highlighted lesions within segmented regions, enhancing model interpretability. In a prospective multi-center in-silico trial with 20 clinicians, AI assistance increased diagnostic accuracy from 0.72 to 0.77 and Cohen's kappa from 0.43 to 0.53, while reducing review time per case by 40%. These results demonstrate that anatomy-aware foundation models with counterfactual explainability can enable accurate, interpretable, and efficient PCa risk assessment, supporting their potential use as virtual biopsies in clinical practice.

Stroke is among the top three causes of death worldwide, and accurate identification of stroke lesion boundaries is critical for diagnosis and treatment. Supervised deep learning methods have emerged as the leading solution for stroke lesion segmentation but require large, diverse, and annotated datasets. The ISLES'24 challenge addresses this need by providing longitudinal stroke imaging data, including CT scans taken on arrival to the hospital and follow-up MRI taken 2-9 days from initial arrival, with annotations derived from follow-up MRI. Importantly, models submitted to the ISLES'24 challenge are evaluated using only CT inputs, requiring prediction of lesion progression that may not be visible in CT scans for segmentation. Our winning solution shows that a carefully designed preprocessing pipeline including deep-learning-based skull stripping and custom intensity windowing is beneficial for accurate segmentation. Combined with a standard large residual nnU-Net architecture for segmentation, this approach achieves a mean test Dice of 28.5 with a standard deviation of 21.27.