We propose a cascaded 3D diffusion model framework to synthesize high-fidelity 3D PET/CT volumes directly from demographic variables, addressing the growing need for realistic digital twins in oncologic imaging, virtual trials, and AI-driven data augmentation. Unlike deterministic phantoms, which rely on predefined anatomical and metabolic templates, our method employs a two-stage generative process. An initial score-based diffusion model synthesizes low-resolution PET/CT volumes from demographic variables alone, providing global anatomical structures and approximate metabolic activity. This is followed by a super-resolution residual diffusion model that refines spatial resolution. Our framework was trained on 18-F FDG PET/CT scans from the AutoPET dataset and evaluated using organ-wise volume and standardized uptake value (SUV) distributions, comparing synthetic and real data between demographic subgroups. The organ-wise comparison demonstrated strong concordance between synthetic and real images. In particular, most deviations in metabolic uptake values remained within 3-5% of the ground truth in subgroup analysis. These findings highlight the potential of cascaded 3D diffusion models to generate anatomically and metabolically accurate PET/CT images, offering a robust alternative to traditional phantoms and enabling scalable, population-informed synthetic imaging for clinical and research applications.

In recent years, scientists have been trying to convert magnetic resonance imaging (MRI) images into computed tomography (CT) images for dose calculations while taking advantage of the benefits of MRI images. The main approaches for image conversion are bulk density, Atlas registration, and machine learning. These methods have limitations in accuracy and time consumption and require large datasets to convert images. In this study, the novel 'voxels spawn voxels' technique combined with the 'orthonormalize' feature in Carimas software was developed to build a conversion dataset from MRI intensity to Hounsfield unit value for some structural regions including gluteus maximus, liver, kidneys, spleen, pancreas, and colon. The original CT images and the converted MRI images were imported into the Geant4/Gamos software for dose calculation. It gives good results (<5%) in most organs except the intestine (18%).

In multi-center neuroimaging studies, the technical variability caused by the batch differences could hinder the ability to aggregate data across sites, and negatively impact the reliability of study-level results. Recent efforts in neuroimaging harmonization have aimed to minimize these technical gaps and reduce technical variability across batches. While Generative Adversarial Networks (GAN) has been a prominent method for addressing harmonization tasks, GAN-harmonized images suffer from artifacts or anatomical distortions. Given the advancements of denoising diffusion probabilistic model which produces high-fidelity images, we have assessed the efficacy of the diffusion model for neuroimaging harmonization. While GAN-based methods intrinsically transform imaging styles between two domains per model, we have demonstrated the diffusion model's superior capability in harmonizing images across multiple domains with single model. Our experiments highlight that the learned domain invariant anatomical condition reinforces the model to accurately preserve the anatomical details while differentiating batch differences at each diffusion step. Our proposed method has been tested using T1-weighted MRI images from two public neuroimaging datasets of ADNI1 and ABIDE II, yielding harmonization results with consistent anatomy preservation and superior FID score compared to the GAN-based methods. We have conducted multiple analyses including extensive quantitative and qualitative evaluations against the baseline models, ablation study showcasing the benefits of the learned domain invariant conditions, and improvements in the consistency of perivascular spaces segmentation analysis and volumetric analysis through harmonization.

The fetal standard plane acquisition task aims to detect an Ultrasound (US) image characterized by specified anatomical landmarks and appearance for assessing fetal growth. However, in practice, due to variability in human operator skill and possible fetal motion, it can be challenging for a human operator to acquire a satisfactory standard plane. To support a human operator with this task, this paper first describes an approach to automatically predict the fetal head standard plane from a video segment approaching the standard plane. A transformer-based image predictor is proposed to produce a high-quality standard plane by understanding diverse scales of head anatomy within the US video frame. Because of the visual gap between the video frames and standard plane image, the predictor is equipped with an offset adaptor that performs domain adaption to translate the off-plane structures to the anatomies that would usually appear in a standard plane view. To enhance the anatomical details of the predicted US image, the approach is extended by utilizing a second modality, US probe movement, that provides 3D location information. Quantitative and qualitative studies conducted on two different head biometry planes demonstrate that the proposed US image predictor produces clinically plausible standard planes with superior performance to comparative published methods. The results of dual-modality solution show an improved visualization with enhanced anatomical details of the predicted US image. Clinical evaluations are also conducted to demonstrate the consistency between the predicted echo textures and the expected echo patterns seen in a typical real standard plane, which indicates its clinical feasibility for improving the standard plane acquisition process.

Advanced autoregressive models have reshaped multimodal AI. However, their transformative potential in medical imaging remains largely untapped due to the absence of a unified visual tokenizer -- one capable of capturing fine-grained visual structures for faithful image reconstruction and realistic image synthesis, as well as rich semantics for accurate diagnosis and image interpretation. To this end, we present MedITok, the first unified tokenizer tailored for medical images, encoding both low-level structural details and high-level clinical semantics within a unified latent space. To balance these competing objectives, we introduce a novel two-stage training framework: a visual representation alignment stage that cold-starts the tokenizer reconstruction learning with a visual semantic constraint, followed by a textual semantic representation alignment stage that infuses detailed clinical semantics into the latent space. Trained on the meticulously collected large-scale dataset with over 30 million medical images and 2 million image-caption pairs, MedITok achieves state-of-the-art performance on more than 30 datasets across 9 imaging modalities and 4 different tasks. By providing a unified token space for autoregressive modeling, MedITok supports a wide range of tasks in clinical diagnostics and generative healthcare applications. Model and code will be made publicly available at: https://github.com/Masaaki-75/meditok.

Background and purposeMRI is critical for the detection of subtle cortical pathology in epilepsy surgery assessment. This can be aided by improved MRI quality and resolution using ultra-high field (7T). But poor access and long scan durations limit widespread use, particularly in a paediatric setting. AI-based learning approaches may provide similar information by enhancing data obtained with conventional MRI (3T). We used a convolutional neural network trained on matched 3T and 7T images to enhance quantitative R1-maps (longitudinal relaxation rate) obtained at 3T in paediatric epilepsy patients and to determine their potential clinical value for lesion identification. Materials and MethodsA 3D U-Net was trained using paired patches from 3T and 7T R1-maps from n=10 healthy volunteers. The trained network was applied to enhance paediatric focal epilepsy 3T R1 images from a different scanner/site (n=17 MRI lesion positive / n=14 MR-negative). Radiological review assessed image quality, as well as lesion identification and visualization of enhanced maps in comparison to the 3T R1-maps without clinical information. Lesion appearance was then compared to 3D-FLAIR. ResultsAI enhanced R1 maps were superior in terms of image quality in comparison to the original 3T R1 maps, while preserving and enhancing the visibility of lesions. After exclusion of 5/31 patients (due to movement artefact or incomplete data), lesions were detected in AI Enhanced R1 maps for 14/15 (93%) MR-positive and 4/11 (36%) MR-negative patients. ConclusionAI enhanced R1 maps improved the visibility of lesions in MR positive patients, as well as providing higher sensitivity in the MR-negative group compared to either the original 3T R1-maps or 3D-FLAIR. This provides promising initial evidence that 3T quantitative maps can outperform conventional 3T imaging via enhancement by an AI model trained on 7T MRI data, without the need for pathology-specific information.

Synthetic CT (sCT) generation from cone-beam CT (CBCT) must maintain stable performance and allow for accurate dose calculation across all treatment fractions to effectively support adaptive proton therapy. This study evaluated a 3D deep-learning (DL) network for sCT generation for prostate cancer patients over the full treatment course. Patient data from 25/6 prostate cancer patients were used to train/test the DL network. Patients in the test set had a planning CT, 39 CBCT images, and at least one repeat CT (reCT) used for replanning. The generated sCT images were compared to fan-beam planning and reCT images in terms of i) CT number accuracy and stability within spherical regions-of-interest (ROIs) in the bladder, prostate, and femoral heads, ii) proton range calculation accuracy through single-spot plans, and iii) dose trends in target coverage over the treatment course (one patient). The sCT images demonstrated image quality comparable to CT, while preserving the CBCT anatomy. The mean CT numbers on the sCT and CT images were comparable, e.g. for the prostate ROI they ranged from 29 HU to 59 HU for sCT, and from 36 HU to 50 HU for CT. The largest median proton range difference was 1.9 mm. Proton dose calculations showed excellent target coverage (V95%≥99.6 %) for the high-dose target. The DL network effectively generated high-quality sCT images with CT numbers, proton range, and dose characteristics comparable to fan-beam CT. Its robustness against intra-patient variations makes it a feasible tool for adaptive proton therapy.

Osteoporotic vertebral compression fractures (OVCFs) are prevalent in the elderly population, typically assessed on computed tomography (CT) scans by evaluating vertebral height loss. This assessment helps determine the fracture's impact on spinal stability and the need for surgical intervention. However, the absence of pre-fracture CT scans and standardized vertebral references leads to measurement errors and inter-observer variability, while irregular compression patterns further challenge the precise grading of fracture severity. While deep learning methods have shown promise in aiding OVCFs screening, they often lack interpretability and sufficient sensitivity, limiting their clinical applicability. To address these challenges, we introduce a novel vertebra synthesis-height loss quantification-OVCFs grading framework. Our proposed model, HealthiVert-GAN, utilizes a coarse-to-fine synthesis network designed to generate pseudo-healthy vertebral images that simulate the pre-fracture state of fractured vertebrae. This model integrates three auxiliary modules that leverage the morphology and height information of adjacent healthy vertebrae to ensure anatomical consistency. Additionally, we introduce the Relative Height Loss of Vertebrae (RHLV) as a quantification metric, which divides each vertebra into three sections to measure height loss between pre-fracture and post-fracture states, followed by fracture severity classification using a Support Vector Machine (SVM). Our approach achieves state-of-the-art classification performance on both the Verse2019 dataset and in-house dataset, and it provides cross-sectional distribution maps of vertebral height loss. This practical tool enhances diagnostic accuracy in clinical settings and assisting in surgical decision-making.

This study systematically examines the impact of training database size and the generalizability of deep learning models for synthetic medical image generation. Specifically, we employ a Cycle-Consistency Generative Adversarial Network (CycleGAN) with softly paired data to synthesize kilovoltage computed tomography (kVCT) images from megavoltage computed tomography (MVCT) scans. Unlike previous works, which were constrained by limited data availability, our study uses an extensive database comprising 4,000 patient CT scans, an order of magnitude larger than prior research, allowing for a more rigorous assessment of database size in medical image translation. We quantitatively evaluate the fidelity of the generated synthetic images using established image similarity metrics, including Mean Absolute Error (MAE) and Structural Similarity Index Measure (SSIM). Beyond assessing image quality, we investigate the model's capacity for generalization by analyzing its performance across diverse patient subgroups, considering factors such as sex, age, and anatomical region. This approach enables a more granular understanding of how dataset composition influences model robustness.

Dynamic FDG PET imaging study of n = 52 rats including 26 control Wistar-Kyoto (WKY) rats and 26 experimental spontaneously hypertensive rats (SHR) were performed using a Siemens microPET and Albira trimodal scanner longitudinally at 1, 2, 3, 5, 9, 12 and 18 months of age. A 15-parameter dual output model correcting for spill over contamination and partial volume effects with peak fitting cost functions was developed for simultaneous estimation of model corrected blood input function (MCIF) and kinetic rate constants for dynamic FDG PET images of rat heart in vivo. Major drawbacks of this model are its dependence on manual annotations for the Image Derived Input Function (IDIF) and manual determination of crucial model parameters to compute MCIF. To overcome these limitations, we performed semi-automated segmentation and then formulated a Long-Short-Term Memory (LSTM) cell network to train and predict MCIF in test data using a concatenation of IDIFs and myocardial inputs and compared them with reference-modeled MCIF. Thresholding along 2D plane slices with two thresholds, with T1 representing high-intensity myocardium, and T2 representing lower-intensity rings, was used to segment the area of the LV blood pool. The resultant IDIF and myocardial TACs were used to compute the corresponding reference (model) MCIF for all data sets. The segmented IDIF and the myocardium formed the input for the LSTM network. A k-fold cross validation structure with a 33:8:11 split and 5 folds was utilized to create the model and evaluate the performance of the LSTM network for all datasets. To overcome the sparseness of data as time steps increase, midpoint interpolation was utilized to increase the density of datapoints beyond time = 10 minutes. The model utilizing midpoint interpolation was able to achieve a 56.4% improvement over previous Mean Squared Error (MSE).