Journal of Medical Imaging

Objectives To assess the effectiveness of our proposed urinary segmentation method based upon image processing, and to contrast its performance against various approaches. Subjects and Methods Non-contrast-enhanced computed tomography (NCCT) scans of instances with stone disease were collected. An experienced urologist and a senior radiology resident generated ground truths by manually drawing the regions of interest (ROIs) of stone samples. The proposed method took loosely defined ROIs as input and included three steps: (1) finding mean_HU_rim, the mean Hounsfield unit (HU) of the stone rim, on each NCCT slice by an edge detection operator; (2) considering pixels with HU higher than the minimum value of all the mean_HU_rim; and (3) removing noise. The entire data was divided into a developing set of 406 samples to evaluate the performance of the proposed method as well as to fine-tune a nnUNetv2 model and a comparison test set of 125 samples to compare the performance and inference time of various segmentation methods. In addition, the reproducibility of our method was tested on the 30 samples randomly selected from the entire data. Results A total of 531 stone samples from 287 instances were included. Our method showed a high level of agreement with the ground truth in the developing set, with the median (interquartile range) Dice coefficient of 0.86 (0.81 - 0.89). The reproducibility of our method was remarkably high with the median (interquartile range) Dice coefficient of 1.0 (0.991 - 1.0). On the comparison test set, our semi-automatic segmentation outperformed the fixed thresholding and deep-learning-based methods. Lastly, the inference time of the proposed method was less than a second. Conclusion The proposed semi-automatic approach provided a consistently reliable, simple, and robust method to segment urinary stones.

Background and purpose: Accurate identification of cardiac MRI volume orientations is essential for reliable image interpretation and for enabling downstream automated analysis pipelines. However, orientation labels and commonly assigned manually, making the process time-consuming and prone to variability. Recent advances in vision-language models offers new opportunities for automated label generation by jointly modeling visual and semantic information. Methods: In this study, we systematically compare a conventional end-to-end transformer-based captioning model with a CLIPCap architecture for automated cardiac MRI view labelling. Both models were evaluated on a dataset of 1184 de-identified cardiac MR images spanning across six clinically relevant orientations: short axis, two-, three-, four-chamber views, left and right ventricular outflow tract views. The caption transformer was trained from scratch, whereas CLIPCap leveraged pretrained CLIP image embeddings combined with lightweight mapping network and a frozen language model. Results: Experimental results demonstrate a substantial performance advantage from CLIPCap, which achieved an overall accuracy of 98% compared to 68% for the caption transformer. CLIPCap consistently delivered high precision and recall across all orientation classes, including those of high clinical importance, while the end to end transformer exhibited unstable performance and complete failure in certain views. Conclusion: These findings highlight the benefits of leveraging large scale multimodal pretrained model for medical image labeling tasks, particularly in data limited settings. The study suggests that CLIPCap provides a more robust and clinically reliable solution for automated cardiac MRI orientation labeling, supporting its integration into efficient and scalable clinical workflows.

Purpose: To evaluate whether visually assessed high-quality prostate MRI, as defined by the Prostate Imaging Quality (PI-QUAL) scoring system, is sufficient to ensure consistent AI model performance, or whether hidden acquisition-related technical effects introduce variability that cannot be detected through visual inspection. Methods: This retrospective study included 568 bpMRI scans from seven institutions, classified using PI-QUAL v2 into high-quality (scores 2-3) and low-quality (score 1) groups. Quantitative image quality features were extracted using MRQy from T2w and ADC sequences. Three experiments were conducted on high-quality scans: (1) unsupervised clustering to identify hidden technical batch effects; (2) V-Net models (M) for prostate capsule segmentation on T2w using single-cluster and balanced multi-cluster training strategies; and (3) VoxelMorph-based deformable registration (M2) to align ADC maps to T2w images. Performance metrics included DSC, precision, recall, volume ratio, HD95, ASSD, and TRE. Kruskal-Wallis tests assessed cross-cluster variability, Two One-Sided Tests (TOST) evaluated performance equivalence, and bootstrap resampling quantified training efficiency. Results: Clustering identified four clusters among 277 high-quality bpMRI scans. Single-cluster models showed variable performance (DSC=0.41-0.85) with significant batch effects (p<0.001), even with 140 training samples. The balanced multi-cluster model (n=60) eliminated batch effects (p=0.075), achieved equivalent performance (p<0.001), and demonstrated 2.23-fold greater training efficiency despite using 2.3-fold fewer samples. Registration performance also varied significantly across clusters (TRE=0.86-5.47 mm, p<0.001). Conclusion: Acquisition-related batch effects significantly impact AI generalizability in prostate MRI, even among high-quality scans. Quantitative clustering enables detection of technical variations and supports diversity-aware training strategies that achieve equivalent performance with 2.3-fold fewer samples while eliminating batch effects.

Faster ultrahigh-resolution (UHR) photon-counting CT acquisitions facilitate breath-hold imaging, but identical reconstruction settings may conceal differences in spatial resolution. To assess spatial resolution and depiction of small airways and venous stents across helical UHR acquisition configurations. Stationary wire, airway, and venous-stent phantoms underwent one acquisition per configuration using five configurations on a dual-source scanner at 120 kVp. Reconstructions used 0.2- and 0.4-mm thicknesses, the same sharp kernel, and iterative reconstruction strength 1. We measured the full-width at half-maximum of the slice sensitivity profile (SSP) and spatial frequency at 20% of the modulation transfer function (MTF20). Rank-based F tests with Holm adjustment assessed configuration differences in resolution and variability across slices. Airway detection used a profile contrast-to-noise criterion; stent assessment was qualitative. Calculated acquisition time for 400-mm coverage decreased from 9.80 to 1.30 seconds between the slowest and fastest configurations. At 0.2-mm reconstruction thickness, median SSP increased from 0.316 to 0.496 mm, and median MTF20 decreased from 2.870 to 2.059 lp/mm (both P<.001). Interquartile ranges increased from 0.047 to 0.197 mm for SSP and from 0.256 to 0.654 lp/mm for MTF20 (both variability P<.001). Increasing reconstruction thickness to 0.4 mm increased SSP by a median 0.221 mm across 2459 matched slice-wire measurements (P<.001). At 0.2-mm reconstruction thickness, 0.4- and 0.6-mm lumens with 0.2-mm walls were detected only in the slowest configuration. Faster configurations showed less distinct stent struts and greater adjacent-wall distortion. Faster helical UHR photon-counting CT configurations had poorer spatial resolution despite identical nominal slice thickness. Consistent acquisition and reconstruction settings are important for longitudinal comparisons.

Accurate breast lesion segmentation in ultrasound is difficult because speckle noise, heterogeneous lesion appearance, and poorly defined margins can obscure clinically relevant boundaries. This work presents UGB-SegNet, an uncertainty-guided breast ultrasound segmentation network that combines four components: multi-scale convolutional block attention (CBAM) at three EfficientNet-B0 encoder stages, a learnable feature pyramid network for adaptive multi-scale fusion, an entropy-guided decoder that routes fine-grained encoder information toward uncertain regions, and a boundary-aware composite loss with deep supervision. The model was evaluated on the public BUSI dataset (647 annotated lesion images) using a stratified 454/64/129 train/validation/test split. On the held-out test set, UGB-SegNet achieved 81.30% Dice, 68.49% IoU, 81.71% sensitivity, 98.14% specificity, and 15.63-pixel HD95. Under identical training conditions it outperformed five directly trained baselines: U-Net, Attention U-Net, SegNet, DeepLabV3+, and TransUNet. A five-fold resampling experiment over the development set with the held-out test set fixed yielded a mean Dice of 76.36% +/- 1.66% (95% CI 74.90-77.81). Reported ablations attribute gains of +2.06 Dice points to multi-scale CBAM, +1.52 to learnable FPN fusion, +1.22 to the uncertainty-guided decoder, and +1.24 to boundary-aware training. UGB-SegNet contains 14.00M parameters and runs at 62.4 FPS at 224 x 224 resolution on the tested NVIDIA GPU, providing a compact accuracy-efficiency trade-off for future prospective clinical validation.

Background and Objective: Micro-ultrasound (MicroUS) is an emerging high-resolution imaging modality for prostate cancer diagnosis and real-time biopsy guidance, but its automated analysis is limited by labor-intensive expert annotation. We aimed to develop and evaluate a foundation model for data-efficient MicroUS image analysis. Methods: We developed muFM, a MicroUS-specific foundation model pretrained using DINOv2-style self-supervised learning on 1,736,391 unlabeled images from 780 patients. The model was fine-tuned and evaluated in three downstream cohorts for prostate capsule segmentation (public, 75 patients), urethra segmentation (institutional, 105 patients), and prostate cancer lesion localization (institutional, 190 patients). Under limited annotation, muFM was compared with randomly initialized models and foundation models pretrained on natural, medical, and ultrasound images. Evaluation used the Dice scores and 95th-percentile Hausdorff distance (HD95) for segmentation and free-response receiver operating characteristic analysis for lesion localization. Key Findings and Limitations: muFM improved performance across all three tasks, with the largest gains under limited annotation. With 10% of training scans, muFM achieved Dice scores of 93.8% for prostate capsule segmentation and 58.4% for urethra segmentation, versus 87.9% and 47.0% for the strongest comparators. Corresponding HD95 values were 2.0 and 4.3 mm. For lesion localization, muFM outperformed other models with approximately 0.75 sensitivity at two false-positive detections per 3D volume. Limitations include retrospective single-institution pretraining and lack of explicit 3D modeling. Conclusions and Clinical Implications: MicroUS-specific self-supervised pretraining improved data-efficient image analysis across prostate and urethra segmentation, and lesion localization. This approach may reduce annotation requirements for developing automated tools in MicroUS image analysis.

To compare virtual noncontrast (VNC) error between deep-silicon photon-counting CT (PCCT) and dual-energy energy-integrating detector CT (EID-CT) in a matched phantom and characterize tissue-specific error in humans. This prospective exploratory study included a matched phantom experiment and pilot human evaluation. Prototype projection-domain water-density maps served as the VNC surrogate. Twenty-four vials containing 0.1--20 mg I/mL in deionized water or simulated blood were scanned with PCCT and EID-CT. Three men underwent true noncontrast and contrast-enhanced PCCT; 80 matched measurements were analyzed. Paired t tests and mixed-effects models assessed system differences and iodine associations. Three participants (mean age, 67 {+/-} 10 years; all men) were included. Mean VNC error was 29.1 HU lower with PCCT than EID-CT (95% CI, 15.5--42.7 HU; {rho} < 0.001). Error was not associated with iodine concentration for PCCT (slope, 0.03 HU per mg I/mL; 95% CI, -0.92 to 0.98; {rho} = 0.95) but increased for EID-CT (slope, 3.51 HU per mg I/mL; 95% CI, 2.56--4.46 HU; {rho} < 0.001). The slopes differed by 3.48 HU per mg I/mL ({rho} < 0.001). In humans, errors in five non-blood-pool tissues were within 5 HU. Error was not associated with iodine concentration among 61 non-blood-pool measurements (adjusted slope, -0.17 HU per mg I/mL; 95% CI, -1.22 to 0.88; {rho} = 0.75). PCCT provided VNC measurements with errors closer to zero and less dependent on iodine than EID-CT in the phantom. In humans, error was generally small in non-blood-pool tissues and was not associated with iodine concentration.

Head and neck radiation oncology, with its complex locality and epidemiology, requires various imaging techniques to achieve sufficient visualization. Through collaboration with medical physicists, radiation therapists, radiation oncologists, and neuroradiologists, our group clinically implemented a standard set of magnetic resonance imaging (MRI) sequences on an MR-simulation scanner for three anatomical regions: skull base, oral cavity, and larynx/pharynx. A total of 41 patients with head and neck cancer and 3 volunteers were scanned on either a 1.5T or 3T MR-simulation scanner. Sequences described include Dixon-based approaches, SPAIR-based approaches, short T2-based approaches, and quantitative imaging methods. We describe and outline the steps to optimize these notable MRI sequences to aid in successful implementation in a radiation oncology clinic.

Objectives: Intraoperative ultrasound (IOUS) is an increasingly valuable real-time imaging modality in neurosurgery, enabling visualization of brain parenchyma, tumor margins, ventricular anatomy, and vascular structures during cranial procedures. Despite its clinical relevance, the development of artificial intelligence (AI)-assisted interpretation tools for neurosurgical IOUS has been limited by the scarcity of large, annotated datasets. Manual pixel-level annotation of ultrasound video frames is labor-intensive, requires specialized neuroanatomical expertise, and scales poorly. To address this bottleneck, we developed a proof-of-concept interactive semi-automated annotation pipeline designed to explore the feasibility of scalable, expert-guided labeling for intraoperative neurosurgical IOUS. Methods: We implemented the system as a napari-based graphical interface coupled with a scikit-learn Random Forest (RF) pixel classifier. Real intraoperative neurosurgical ultrasound videos from five craniotomy cases were used as training material. A neurosurgical expert annotated a minimal set of representative frames using brush-based labeling to define anatomical classes of interest. For each labeled pixel, a feature vector was computed comprising raw intensity, multi-scale Gaussian filtering ({sigma} = 1, 2, 3), Laplacian of Gaussian, and Sobel edge magnitude. The RF classifier was trained on accumulated annotations using class-balanced weighting and out-of-bag (OOB) validation, then applied across full video sequences. An incremental accumulator module enabled multi-case training. Segmentation outputs were exported as TIFF overlays for expert quality review. Results: Across five intraoperative cases, sparse expert annotation of a small number of representative frames per case was sufficient to train case-specific Random Forest classifiers that propagated labels across the full video sequences. Out-of-bag estimation served as an internal coherence check during training, and the classifier additionally produced per-class probability maps that exposed regions of uncertainty at acoustic boundaries. On expert qualitative review, the auto-segmented frames were anatomically coherent, with the classifier reliably distinguishing anechoic fluid compartments from echogenic tissue interfaces; residual misclassification was concentrated in low-contrast boundary regions. Conclusions: This proof-of-concept study demonstrates that semi-automated, expert-guided annotation of intraoperative neurosurgical ultrasound is technically feasible using classical machine learning. By combining neurosurgical domain expertise with an interactive, incremental annotation workflow applied to real intraoperative acquisitions, this framework represents a foundational step toward scalable generation of labeled datasets for neurosurgical IOUS. Future directions include formal multi-rater validation, expanded multi-class anatomical labeling, and eventual migration to deep learning architectures as annotated datasets grow.

Deep-silicon photon-counting CT uses a detector architecture distinct from cadmium-telluride systems and has not been evaluated for pediatric head imaging. To compare attenuation-dependent performance of deep-silicon photon-counting CT (dSi-PCCT) with energy-integrating detector CT (EID-CT) and assess spectral and ultra-high-resolution (UHR) reconstructions from one acquisition. Graded-thickness cranial attenuator and pediatric head phantoms were scanned at a matched target volume CT dose index of 18 mGy. dSi-PCCT used 120 kVp and generated 50-, 65-, and 100-keV images; EID-CT used 80 and 100 kVp. We measured CT-number bias, noise, contrast, and contrast-to-noise ratio (CNR) across 0--9 mm of nominal skull thickness. Mixed-effects models assessed imaging-condition effects. We compared matched standard-resolution and UHR temporal-bone images and generated calcium- and water-density maps from the same dSi-PCCT scan. Noise increased by 4.54%, 3.07%, and 1.35% per millimeter at 50, 65, and 100 keV, respectively, versus 8.38% and 7.84% at 80 and 100 kVp. Corresponding CNR decreases were 4.17%, 3.25%, and 1.75% versus 8.94% and 8.54% (interaction {rho} < 0.001). Changes in absolute CT-number bias and contrast did not differ among conditions ({rho} = 0.92 and {rho} = 0.59). In the pediatric head phantom, model-estimated 50-keV CNR ratios were 1.299 (95% CI, 1.163--1.452) versus 80 kVp and 1.424 (95% CI, 1.275--1.592) versus 100 kVp. Across three temporal-bone edges, UHR transition widths were 58.6%-61.9% narrower than standard-resolution widths. dSi-PCCT maintained CNR more consistently as nominal skull thickness increased and provided complementary spectral and UHR pediatric head images from one acquisition.

Purpose: To quantify the sampling error of conventional 2D regions of interest (ROIs) compared with whole-organ 3D volumetric CT attenuation for opportunistic hepatic steatosis screening, and to evaluate its dependency on intraparenchymal spatial heterogeneity. Materials and Methods: In this retrospective computational study (CLAIM-compliant), 971 abdominal CT scans were analyzed across a multicenter derivation cohort (TotalSegmentator v2, n = 841) and an external validation cohort (Medical Segmentation Decathlon Task03, n = 130). Automated 3D liver segmentation was performed using nnU-Net, followed by an 8-mm spherical 3D erosion to eliminate peripheral partial volume effects and major hilar vessels. Three clinical sampling paradigms (1, 3, and 5 axial 2D ROIs of 15 mm diameter) were modeled through 500 stochastic Monte Carlo iterations per scan (485,500 total simulations). Absolute sampling error (|Delta HU| = |HU_ROI - HU_3D|) and diagnostic misclassification rates at the standard <= 40 HU steatosis threshold were computed. Results: In the derivation cohort, the conventional 3-ROI approach yielded a mean absolute error of 6.61 +/- 7.83 HU (median: 2.92 HU), significantly underperforming compared to the 5-ROI strategy (5.21 +/- 6.00 HU; Wilcoxon paired test, p < 0.001) but improving on the single-ROI approach (11.32 +/- 14.16 HU; p < 0.001). Sampling error correlated strongly with parenchymal spatial heterogeneity (Pearson r = 0.850, 95% CI: [0.830-0.868], p < 0.001). In borderline attenuation zones (35-45 HU), diagnostic misclassification reached 45.2% in heterogeneous livers. Findings were corroborated on external validation, confirming the reproducible decay of sampling error (|Delta HU| = 7.38 HU for 1 ROI vs. 5.14 HU for 3 ROI, p < 0.001). Conclusion: Manual 2D ROI sampling exhibits an intrinsic metrological uncertainty exceeding clinically relevant thresholds (5 HU) in heterogeneous livers, misclassifying nearly half of borderline cases. Opportunistic screening requires fully automated 3D volumetric extraction embedded into clinical PACS workflows.

Trans-arterial radioembolization using yttrium-90 (Y-90) microspheres treats unresectable liver cancer with radiation. The tumor-to-parenchyma dose ratio is mainly governed by the patient-specific hepatic arterial flow distribution, which transports the microspheres. Even though computational fluid dynamics (CFD) simulation can predict this flow distribution, its high computational cost and time limit intraprocedural application. To overcome this computational limitation, a physics-informed neural network (PINN) surrogate was developed to provide rapid predictions of flow distribution in a patient-specific hepatic artery model. The surrogate maps spatial coordinates to the velocity vector U with Cartesian components u, v, w and the static pressure p. Reference velocity and pressure fields came from a converged finite-element CFD solve of the same geometry at about 1.05 x 105 nodes; a controlled fraction was used to supervise the training while the remainder was held out for validation. Using the NVIDIA PhysicsNeMo framework, we compared four coordinate networks (MLP, Modified Fourier, Multi-Scale Fourier and SIREN) under soft and hard inlet enforcement across a supervised data sweep (data fraction df = 0 to 0.20) and repeated df = 0.20 at a matched 300,000-step budget. Predicted flow splits were propagated through an established Y-90 dose-kernel pipeline. Accuracy is governed primarily by the supervised-data fraction, not the architecture: at df [&ge;] 0.10 the backbones agree to within one to two R2 points, and R2(u) rises from {approx} 0 under pure physics (df = 0) to {approx} 0.95 by df = 0.01 for the best backbones, saturating near 0.97 to 0.99 by df = 0.20. At a matched 300,000-step budget the Multi-Scale Fourier network leads (R2(u) = 0.978) ahead of SIREN (R2(u) = 0.976) and the MLP (R2(u) = 0.968); the best value seen (R2(u) = 0.985, Modified Fourier) needed 1.8 times that budget. On the df = 0.20 checkpoint the per-outlet flow split matches CFD (r = 0.999, L1 = 4.4 percentage points) and the derived dosimetry agrees to within 2.4% for perfused volume, mean dose and every isodose volume. At df = 0 training collapses toward the trivial near-zero state, which satisfies continuity, momentum and the zero-pressure outlets exactly and is ruled out only by the inlet and data terms; this, not architecture choice, is the limiting failure mode. Once trained, the network evaluates the full 105,284-node field in about 2 s on a four-core CPU. The supervised-data fraction, with a practical floor between df = 0.05 and 0.10, is the primary design lever for such hepatic-arterial PINN surrogates.

Automatic segmentation of medical images could accelerate clinical and research workflows, yet existing tools are rarely compared independently on shared benchmarks, making it difficult to assess genuine progress. This is particularly consequential for limb muscle segmentation in magnetic resonance imaging (MRI), which underpins volumetric analysis, radiomics, and fat fraction quantification used as biomarkers in neuromuscular disease -- applications where segmentation errors can directly affect interpretation. Despite the availability of multiple open-source segmentation tools spanning diverse architectures, no independent head-to-head comparison has appeared in the literature. We address this gap by evaluating eight tools, six of which run independently, and the other two of which are meant to be auxiliary algorithms once muscles have been somehow initially labeled. We evaluated these tools on three-dimensional MRI volumes across three base datasets and then later against derived datasets, assessing both quantitative metrics and qualitative usability. Our results reveal substantial performance variation across methods and datasets, with many methods showing reduced accuracy on pathological cases. Domain-specific models trained on large datasets consistently outperformed foundation models and newer general-purpose architectures. Our results reveal practical trade-offs between accuracy, generalizability, and ease of use that are critical for clinical adoption, while raising concerns for deployment on underrepresented populations and pathologies.

Background: Comprehensive evaluations of lung nodule detection methods remain limited, particularly regarding data from clinical routine, novel photon-counting CT (PCCT) technology, and the comparison of academic and commercial models. Materials and Methods: This retrospective study collected 1860 routine PCCT scans acquired at Hannover Medical School from 2021 to 2024. We compared four deep-learning-based lung nodule detection models, which were developed for conventional CT, namely the publicly available TotalSegmentator, nnDetection, and RadYOLO methods, as well as a commercial lung computer-aided detection (CAD) system. We evaluated the inter-model agreement in an unannotated cohort (n=1699), sensitivity in a smaller cohort of 25 manually annotated scans, and false positives as detections in 136 report-negative scans. Results: In the unannotated cohort, the commercial CAD system detected 7071, TotalSegmentator 4264, nnDetection 5254, and RadYOLO 3122 nodules. Overall, the inter-model agreement was low, and the models produced substantially different candidate nodules. Approximately 1300 nodules were detected by all models, representing 18.7% - 42.8% of each model's detections. 15.4% - 37.4% of a model's detections did not match any other model. The models showed moderate sensitivity (between 0.71, 95%-CI [0.64, 0.77], and 0.83, 95%-CI [0.77, 0.88]), and TotalSegmentator and RadYOLO produced few false positives per scan (both 0.85, 95%-CI [0.70, 1.02]). Conclusion: The lung nodules detected by four deep learning models showed high variation and a lack of consensus on clinical PCCT data. These findings highlight the importance of model benchmarking and local validation when deploying models in real-world clinical workflows.

Foundation models (FMs) have rapidly become dominant in artificial intelligence and are increasingly being adopted in computational pathology. Numerous pathology-specific FMs have been developed and evaluated for a variety of downstream tasks, most commonly using frozen image embeddings with linear probes. While a pathology FM, at least implicitly suggests broad reusability across tasks, the transferability of these models to neurodegenerative disease-related tasks remains largely unexplored. In this study, we evaluated fourteen frozen feature extractors, including general-vision and pathology FMs, across four pathology image classification datasets spanning two specialized neurodegenerative disease immuno-histochemistry (IHC) tasks (Tau neurofibrillary tangle (NFT) and amyloid-{beta} plaque classification) and two cancer-related hematoxylin and eosin (H&E) tasks (the breast tissue BACH dataset and the multi-tissue TIL dataset). The cancer-related H&E datasets represent tasks more closely aligned with the predominant pretraining domain of many current pathology FMs, whereas the neurodegenerative IHC datasets represent a more specialized domain with limited representation in existing FM pretraining cohorts. Frozen embedding linear probes were compared against a conventional supervised ResNet-50 convolutional neural network (CNN) trained directly on image tiles using identical train, validation, and test splits. Across both neurodegenerative IHC datasets, the supervised CNN substantially outperformed all frozen FM linear probes. In contrast, pathology FMs achieved performance comparable to, and in some cases exceeding, the supervised CNN across the two cancer-related H&E datasets, with UNI2-h achieving the highest performance on BACH and several pathology FMs performing on par with the CNN on the larger TIL dataset. Furthermore, a supervised CNN trained using only 1% of the Tau NFT training data (1,938 tiles) still exceeded the performance of the best frozen FM linear probe trained on the complete dataset. Together, these results suggest the transferability of frozen pathology FMs may depend strongly on how well the downstream task is represented by their pretraining domain. These findings demonstrate frozen pathology FMs transfer effectively to the cancer-related H&E tasks evaluated here but may be less effective for specialized neurodegenerative IHC tasks less well represented in current FM pretraining cohorts. Expanding pathology FM pretraining datasets to include a broader range of disease domains and staining modalities may therefore improve transferability to specialized pathology applications. Our findings also highlight the continued importance of conventional supervised learning and motivate future work investigating nonlinear probes and end-to-end FM fine-tuning.

To evaluate whether eight-bin deep-silicon photon-counting CT can quantify iodine and gadolinium separately and distinguish radiopaque embolic beads from vascular contrast in tumor phantoms. In this phantom study, two-input decomposition of scanner-generated iodine and water images was compared with three- and five-material decomposition of five reconstructed energy-bin images in 56 common test vials, including nine simulated-blood preparations. One tantalum and one bismuth vial were also tested with five-material decomposition. A separately calibrated two-input method using scanner-generated 70-keV and iodine images was applied to four static 3D-printed tumor vascular models: iodine-only and gadolinium-only controls and two models with iodine-containing beads and gadolinium-filled channels. Concentration errors were summarized by bias and root mean square error (RMSE). For iodine, RMSE was 0.40, 0.73, and 1.12 mg I/mL with two-input, three-material, and five-material decomposition, respectively; corresponding gadolinium RMSEs were 0.57, 0.55, and 0.46 mg Gd/mL. Two-input biases were +0.01 mg I/mL and +0.29 mg Gd/mL. Five-material decomposition distinguished tantalum and bismuth in the separate test vials. Beads and vascular gadolinium were spatially separated in the tumor models. In the two bead-containing models, gadolinium measured 32.56 {+/-} 8.02 and 37.02 {+/-} 7.93 mg Gd/mL (mean {+/-} spatial SD), compared with the prepared 35.00 mg Gd/mL. Deep-silicon photon-counting CT separately visualized radiopaque embolic beads and quantified vascular gadolinium from a single acquisition. Recalibrated scanner-generated images supported quantitative dual-contrast imaging, while research energy-bin decomposition distinguished additional materials.

Background: Whole-brain segmentation of neonatal low-field MRI is needed for regional volumetric analysis, but established segmentation systems are designed around different field strengths, contrasts, and age ranges. Methods: We first applied complete end-to-end pipelines, including FreeSurfer, FSL-based processing, Infant FreeSurfer, FastSurfer, BrainSuite, SPM12/CAT12, iBEAT v2, the dHCP structural pipeline, and M-CRIB-S, directly to 0.35T neonatal T2-weighted images. Because those systems did not produce a complete 47-region output, we next tested released end-to-end AI segmentation models, including SynthSeg, FastSurfer, FastSurferVINN, and QuickNAT. These were highly sensitive to input format, contrast, voxel geometry, and the high-field training domain. We then decomposed the task into skull stripping, registration, super-resolution, atlas propagation, and tissue refinement and tested the available components individually. Results: No complete end-to-end pipeline or released AI model delivered a usable 47-region segmentation in direct application. Stage-wise testing showed that preprocessing, reconstruction, registration, atlas propagation, and expectation-maximization each solved only part of the problem. Their final assembly produced complete 47-region outputs across the neonatal cohort. Conclusions: Direct application of established pipelines was insufficient for 0.35T neonatal T2-weighted MRI. A stage-wise combination of compatible components provided the workable solution.

Purpose: Videourodynamics (VUDS) are used to assess bladder dysfunction and risk of upper urinary tract deterioration in patients with spina bifida (SB), but interpretation is subject to interrater variability. Machine learning models have previously classified bladder dysfunction severity and predicted incident hydronephrosis using VUDS data. This study evaluated the performance of adapted versions of these models in an independent external cohort. Materials and Methods: VUDS data were collected from SB patients who underwent VUDS at a single institution between 2016 and 2025. Previously developed machine learning frameworks were adapted for implementation in the independent dataset. Three models were utilized: a deep learning convolutional neural network model using pressure-volume data, a deep learning imaging model using fluoroscopic imaging data, and an ensemble model that averaged the risk data from the pressure-volume and fluoroscopic data. The models were evaluated for prediction of incident hydronephrosis and classification of bladder dysfunction severity compared to expert pediatric urologist reviewers. Results: 70 patients were included in the hydronephrosis cohort, of whom 13 (18%) developed incident hydronephrosis. The ensemble model outperformed individual modalities, with a concordance index of 0.77 (pressure-volume only: 0.70, imaging only: 0.72) and an overall AUROC of 0.75 (pressure-volume: 0.67, imaging: 0.74). For bladder dysfunction classification in 95 VUDS studies, the ensemble model achieved 71% accuracy compared to 60% for pressure-volume data alone and 66% for imaging alone, with no substantial disagreements with expert reviewers. Conclusions: The machine learning models that were developed to predict incident hydronephrosis and classify the severity of bladder dysfunction using VUDS data were able to be adapted to an external, independent dataset and perform with similar discrimination and accuracy as previously reported.

Background: Patient-specific dosimetry in targeted radionuclide therapy (TRT) is increasingly supported by quantitative imaging; however, voxel-scale dose estimates cannot resolve cellular and subcellular heterogeneity present within tumor tissue. Conventional cellular dosimetry models often represent cells as regularly packed spheres, which may not capture realistic tumor microarchitecture. This study investigated how histopathology-informed three-dimensional (3D) tumor microarchitecture influences cellular absorbed-dose distributions, predicted radiobiological response, and treatment-related conclusions in TRT. Methods: Realistic 3D cellular tumor models were constructed from thick-section 3D cyclic immunofluorescence (3D CyCIF) imaging data. Monte Carlo radiation-transport simulations were performed for Lu-177, Pd-103, Tb-161 and Y-90, with activity localized in cancer-cell subcellular compartments and in non-cancer targets. Absorbed dose per total cluster decay was calculated individually for each cell nucleus and compared with corresponding simplified spherical tumor models. Cell survival fractions were estimated using the linear-quadratic model, and tumor control probability (TCP) was calculated from the individual cancer-cell survival probabilities. Results: Simplified model overestimated mean cancer-cell nuclear absorbed dose by 9.54% to 14.51% for nuclear localization and by 33.23% to 44.88% for cytoplasmic and membrane localization. In larger tissue models, simplified and realistic geometries produced comparable mean absorbed doses in several cases; however, the realistic models consistently generated broader cellular dose distributions, including low-dose cancer-cell subpopulations not captured by the simplified models. These dose-distribution differences propagated into wider survival-fraction distributions and altered TCP predictions. In the spatially heterogeneous tumor region, the realistic model required 1.11- to 1.33-fold higher activity to reach TCP90 for nuclear localization, and 1.49- to 5.93-fold higher activity for cytoplasmic and membrane localization. For similar tumor control, the radionuclide and source-localization combination associated with the greatest preservation of surrounding non-cancer cells differed between the simplified and realistic models. The intra-voxel analysis further showed that homogeneous voxel-scale activity assumptions can obscure biologically relevant microscopic dose heterogeneity. Conclusion: Native 3D tumor microarchitecture substantially influences cellular absorbed-dose distributions, predicted biological response, and treatment-related conclusions. Histopathology-informed 3D cellular models can be used as a complementary framework for characterizing cellular dose heterogeneity and may improve the evaluation of radionuclides and subcellular targeting strategies in heterogeneous tumor tissue.

Accurate classification of lung adenocarcinoma growth patterns from hematoxylin and eosin (H\&E)-stained whole-slide images is essential for treatment planning and prognosis. We develop and evaluate an exponentially weighted ensemble deep learning approach that combines five architectures (EfficientNet-B3, DeiT3, Swin Transformer, Vision Transformer, ConvNeXt) with weights determined by a softmax transformation of per-model Quadratic Weighted Kappa (QWK) scores. Using 25,545 image patches at 40x magnification from the Dartmouth-Hitchcock dataset, we evaluated performance with stratified 5-fold cross-validation. Individual model performance ranged from Cohen's kappa kappa = 0.5563 (ConvNeXt) to kappa = 0.9561 (EfficientNet-B3). The ensemble achieved 96.52% accuracy and kappa = 0.9648, exceeding the best individual model by delta kappa = +0.0087 with statistical significance (McNemar's test, chi^2 = 50.84, p < 10^-12; bootstrap 95% CI on delta kappa [+0.0056, +0.0117]. Pairwise disagreement rates of 16.6--43.2% between architectures indicate that the models capture different aspects of tissue morphology, and the framework automatically down-weights unreliable components, as illustrated by ConvNeXt receiving only 5.7% ensemble weight despite a single-fold training failure. The approach offers a clinically grounded method for ensemble aggregation in digital pathology.