Accurate preoperative diagnosis of parotid gland tumors (PGTs) is crucial for surgical planning since malignant tumors require more extensive excision. Though fine-needle aspiration biopsy is the diagnostic gold standard, its sensitivity in detecting malignancies is limited. While Deep learning (DL) models based on magnetic resonance imaging (MRI) are common in medicine, they are less studied for parotid gland tumors. This study used a 2.5D imaging approach (Incorporating Inter-Slice Information) to train a DL model to differentiate between benign and malignant PGTs. This retrospective study included 122 parotid tumor patients, using MRI and clinical features to build predictive models. In the traditional model, univariate analysis identified statistically significant features, which were then used in multivariate logistic regression to determine independent predictors. The model was built using four-fold cross-validation. The deep learning model was trained using 2D and 2.5D imaging approaches, with a transformer-based architecture employed for transfer learning. The model's performance was evaluated using the area under the receiver operating characteristic curve (AUC) and confusion matrix metrics. In the traditional model, boundary and peritumoral invasion were identified as independent predictors for PGTs, and the model was constructed based on these features. The model achieved an AUC of 0.79 but demonstrated low sensitivity (0.54). In contrast, the DL model based on 2.5D T2 fat-suppressed images showed superior performance, with an AUC of 0.86 and a sensitivity of 0.78. The 2.5D imaging technique, when integrated with a transformer-based transfer learning model, demonstrates significant efficacy in differentiating between PGTs.

Breast and thyroid cancers are among the most prevalent and fastest growing malignancies worldwide with ultrasound imaging serving as the primary modality for screening and surgical navigation of these lesions. Accurate and real-time lesion segmentation in ultrasound images is crucial for guiding precise needle placement during biopsies and surgeries. To address this clinical need, we propose <b>MLP-UNet</b>, a deep learning model for automatic segmentation of breast tumors and thyroid nodules in ultrasound images. MLP-UNet adopts an encoder-decoder architecture with a U-shaped structure and integrates a MLP-based module(MAP) module within the encoder stage. Attention module is a lightweight employed during the skip connections to enhance feature representation. Using only using 33.75 M parameters, MLP-UNet achieves state-of-the-art segmentation performance. On the BUSI, it attains Dice, IoU, and Recall of 80.61%, 67.93%, and 80.48%, respectively. And on the DDTI, it attains Dice, IoU, and Recall of 81.67% for Dice, 71.72%. These results outperform several classical and state-of-the-art segmentation networks while maintaining low computational complexity, highlighting its significant potential for clinical application in ultrasound-guided surgical navigation systems.

The accurate assessment of computed tomography (CT) image quality is crucial for ensuring diagnostic reliability while minimizing radiation dose. Radiologists' evaluations are time-consuming and labor-intensive. Existing automated approaches often require large CT datasets with predefined image quality assessment (IQA) scores, which often do not align well with clinical evaluations. We aim to develop a reference-free, automated method for CT IQA that closely reflects radiologists' evaluations, reducing the dependency on large annotated datasets. We propose Task-Focused Knowledge Transfer (TFKT), a deep learning-based IQA method leveraging knowledge transfer from task-similar natural image datasets. TFKT incorporates a hybrid convolutional neural network-transformer model, enabling accurate quality predictions by learning from natural image distortions with human-annotated mean opinion scores. The model is pre-trained on natural image datasets and fine-tuned on low-dose computed tomography perceptual image quality assessment data to ensure task-specific adaptability. Extensive evaluations demonstrate that the proposed TFKT method effectively predicts IQA scores aligned with radiologists' assessments on in-domain datasets and generalizes well to out-of-domain clinical pediatric CT exams. The model achieves robust performance without requiring high-dose reference images. Our model is capable of assessing the quality of <math xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mo>∼</mo> <mn>30</mn></mrow> </math> CT image slices in a second. The proposed TFKT approach provides a scalable, accurate, and reference-free solution for CT IQA. The model bridges the gap between traditional and deep learning-based IQA, offering clinically relevant and computationally efficient assessments applicable to real-world clinical settings.

Addressing the critical challenge of precise boundary delineation in medical image segmentation, we introduce DPGNet, an adaptive deep learning model engineered to emulate expert perception of intricate anatomical edges. Our key innovations drive its superior performance and clinical utility, encompassing: 1) a three-stage progressive refinement strategy that establishes global context, performs hierarchical feature enhancement, and precisely delineates local boundaries; 2) a novel Edge Difference Attention (EDA) module that implicitly learns and quantifies boundary uncertainties without requiring explicit ground truth supervision; and 3) a lightweight, transformer-based architecture ensuring an exceptional balance between performance and computational efficiency. Extensive experiments across diverse and challenging medical image datasets demonstrate DPGNet's consistent superiority over state-of-the-art methods, notably achieving this with significantly lower computational overhead (25.51 M parameters). Its exceptional boundary refinement is rigorously validated through comprehensive metrics (Boundary-IoU, HD95) and confirmed by rigorous clinical expert evaluations. Crucially, DPGNet generates an explicit uncertainty boundary map, providing clinicians with actionable insights to identify ambiguous regions, thereby enhancing diagnostic precision and facilitating more accurate clinical segmentation outcomes. Our code is available at: https://github.fangnengwuyou/DPGNet.

The thickness of the diaphragm serves as a crucial biometric indicator, particularly in assessing rehabilitation and respiratory dysfunction. However, measuring diaphragm thickness from ultrasound images mainly depends on manual delineation of the fascia, which is subjective, time-consuming, and sensitive to the inherent speckle noise. In this study, we introduce an edge-aware diffusion segmentation model (ESADiff), which incorporates prior structural knowledge of the fascia to improve the accuracy and reliability of diaphragm thickness measurements in ultrasound imaging. We first apply a diffusion model, guided by annotations, to learn the image features while preserving edge details through an iterative denoising process. Specifically, we design an anisotropic edge-sensitive annotation refinement module that corrects inaccurate labels by integrating Hessian geometric priors with a backtracking shortest-path connection algorithm, further enhancing model accuracy. Moreover, a curvature-aware deformable convolution and edge-prior ranking loss function are proposed to leverage the shape prior knowledge of the fascia, allowing the model to selectively focus on relevant linear structures while mitigating the influence of noise on feature extraction. We evaluated the proposed model on an in-house diaphragm ultrasound dataset, a public calf muscle dataset, and an internal tongue muscle dataset to demonstrate robust generalization. Extensive experimental results demonstrate that our method achieves finer fascia segmentation and significantly improves the accuracy of thickness measurements compared to other state-of-the-art techniques, highlighting its potential for clinical applications.

Extrapolation plays a critical role in machine/deep learning (ML/DL), enabling models to predict data points beyond their training constraints, particularly useful in scenarios deviating significantly from training conditions. This article addresses the limitations of current convolutional neural networks (CNNs) in extrapolation tasks within image restoration and compressed sensing (CS). While CNNs show potential in tasks such as image outpainting and CS, traditional convolutions are limited by their reliance on interpolation, failing to fully capture the dependencies needed for predicting values outside the known data. This work proposes an extrapolation convolution (EC) framework that models missing data prediction as an extrapolation problem using linear prediction within DL architectures. The approach is applied in two domains: first, image outpainting, where EC in encoder-decoder (EnDec) networks replaces conventional interpolation methods to reduce artifacts and enhance fine detail representation; second, Fourier-based CS-magnetic resonance imaging (CS-MRI), where it predicts high-frequency signal values from undersampled measurements in the frequency domain, improving reconstruction quality and preserving subtle structural details at high acceleration factors. Comparative experiments demonstrate that the proposed EC-DecNet and FDRN outperform traditional CNN-based models, achieving high-quality image reconstruction with finer details, as shown by improved peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and kernel inception distance (KID)/Frechet inception distance (FID) scores. Ablation studies and analysis highlight the effectiveness of larger kernel sizes and multilevel semi-supervised learning in FDRN for enhancing extrapolation accuracy in the frequency domain.

Accurately imaging the spatial distribution of longitudinal speed of sound (SoS) has a profound impact on image quality and the diagnostic value of ultrasound. Knowledge of SoS distribution allows effective aberration correction to improve image quality. SoS imaging also provides a new contrast mechanism to facilitate disease diagnosis. However, SoS imaging is challenging in the pulse-echo mode. Deep learning (DL) is a promising approach for pulse-echo SoS imaging, which may yield more accurate results than pure physics-based approaches. Herein, we developed a robust DL approach for SoS imaging that learns the nonlinear mapping between measured time shifts and the underlying SoS without subjecting to the constraints of a specific forward model. Various strategies were adopted to enhance model performance. Time-shift maps were computed by adopting a common mid-angle configuration from the non-DL literature, normalizing complex beamformed ultrasound data, and accounting for depth-dependent frequency when converting phase shifts to time shifts. The structural similarity index measure (SSIM) was incorporated into the loss function to learn the global structure for SoS imaging. A two-stage training strategy was employed, leveraging computationally efficient ray-tracing synthesis for extensive pretraining, and more realistic but computationally expensive full-wave simulations for fine-tuning. Using these combined strategies, our model was shown to be robust and generalizable across different conditions. The simulation-trained model successfully reconstructed the SoS maps of phantoms using experimental data. Compared with the physics-based inversion approach, our method improved reconstruction accuracy and contrast-to-noise ratio in phantom experiments. These results demonstrated the accuracy and robustness of our approach.

To compare PET-derived metrics between digital and analogue PET/CT in hyperparathyroidism, and to assess whether machine learning (ML) applied to quantitative PET parameters can distinguish parathyroid adenoma (PA) from hyperplasia (PH). From an initial multi-centre cohort of 179 patients, 86 were included, comprising 89 PET-positive lesions confirmed histologically (74 PA, 15 PH). Quantitative PET parameters-maximum standardised uptake value (SUVmax), metabolic tumour volume (MTV), target-to-background ratio (TBR), and maximum diameter-along with serum PTH and calcium levels, were compared between digital and analogue PET scanners using the Mann-Whitney U test. Receiver operating characteristic (ROC) analysis identified optimal threshold values. ML models (LASSO, decision tree, Gaussian naïve Bayes) were trained on harmonised quantitative features to distinguish PA from PH. Digital PET detected significantly smaller lesions than analogue PET, in both metabolic volume (1.32 ± 1.39 vs. 2.36 ± 2.01 cc; p < 0.001) and maximum diameter (8.35 ± 4.32 vs. 11.87 ± 5.29 mm; p < 0.001). PA lesions showed significantly higher SUVmax and TBR compared to PH (SUVmax: 8.58 ± 3.70 vs. 5.27 ± 2.34; TBR: 14.67 ± 6.99 vs. 8.82 ± 5.90; both p < 0.001). The optimal thresholds for identifying PA were SUVmax > 5.89 and TBR > 11.5. The best ML model (LASSO) achieved an AUC of 0.811, with 79.7% accuracy and balanced sensitivity and specificity. Digital PET outperforms analogue system in detecting small parathyroid lesions. Additionally, ML analysis of PET-derived metrics and PTH may support non-invasive distinction between adenoma and hyperplasia.

Development and verification of a convolutional neural network (CNN)-based deep learning (DL) model for mandibular canal (MC) localization on multicenter cone beam computed tomography (CBCT) images. In this study, a total 1056 CBCT scans in multiple centers were collected. Of these, 836 CBCT scans of one manufacturer were used for development of CNN model (training set: validation set: internal testing set = 640:360:36) and an external testing dataset of 220 CBCT scans from other four manufacturers were tested. The convolution module was built using a stack of Conv + InstanceNorm + LeakyReLU. Average symmetric surface distance (ASSD) and symmetric mean curve distance (SMCD) were used for quantitative evaluation of this model for both internal testing data and partial external testing data. Visual scoring (1-5 points) were performed to evaluate the accuracy and generalizability of MC localization for all external testing data. The differences of ASSD, SMCD and visual scores among the four manufacturers were compared for external testing dataset. The time of manual and automatic MC localization were recorded. For the internal testing dataset, the average ASSD and SMCD was 0.486 mm and 0.298 mm respectively. For the external testing dataset, 86.8% CBCT scans' visual scores ≥ 4 points; the average ASSD and SMCD of 40 CBCT scans with visual scores ≥ 4 points were 0.438 mm and 0.185 mm respectively; there were significant differences among the four manufacturers for ASSD, SMCD and visual scores (p < 0.05). And the time for bilateral automatic MC localization was 8.52s (± 0.97s). In this study, a CNN model was developed for automatic MC localization, and external testing of large sample on multicenter CBCT images showed its excellent clinical application potential.

Multimodal data integration has emerged as a transformative approach in the health care sector, systematically combining complementary biological and clinical data sources such as genomics, medical imaging, electronic health records, and wearable device outputs. This approach provides a multidimensional perspective of patient health that enhances the diagnosis, treatment, and management of various medical conditions. This viewpoint presents an overview of the current state of multimodal integration in health care, spanning clinical applications, current challenges, and future directions. We focus primarily on its applications across different disease domains, particularly in oncology and ophthalmology. Other diseases are briefly discussed due to the few available literature. In oncology, the integration of multimodal data enables more precise tumor characterization and personalized treatment plans. Multimodal fusion demonstrates accurate prediction of anti-human epidermal growth factor receptor 2 therapy response (area under the curve=0.91). In ophthalmology, multimodal integration through the combination of genetic and imaging data facilitates the early diagnosis of retinal diseases. However, substantial challenges remain regarding data standardization, model deployment, and model interpretability. We also highlight the future directions of multimodal integration, including its expanded disease applications, such as neurological and otolaryngological diseases, and the trend toward large-scale multimodal models, which enhance accuracy. Overall, the innovative potential of multimodal integration is expected to further revolutionize the health care industry, providing more comprehensive and personalized solutions for disease management.