Journal of Medical Imaging最新文献

Purpose: Four-dimensional (4D) ultrasound imaging is widely used in clinics for diagnostics and therapy guidance. Accurate target tracking in 4D ultrasound is crucial for autonomous therapy guidance systems, such as radiotherapy, where precise tumor localization ensures effective treatment. Supervised deep learning approaches rely on reliable ground truth, making accurate labels essential. We investigate the reliability of expert-labeled ground truth data by evaluating intra- and inter-observer variability in landmark labeling for 4D ultrasound imaging in the liver.

Approach: Eight 4D liver ultrasound sequences were labeled by eight expert observers, each labeling eight landmarks three times. Intra- and inter-observer variability was quantified, and observer survey and motion analysis were conducted to determine factors influencing labeling accuracy, such as ultrasound artifacts and motion amplitude.

Results: The mean intra-observer variability ranged from $1.58\mathrm{mm}\pm 0.90\mathrm{mm}$ to $2.05\mathrm{mm}\pm 1.22\mathrm{mm}$ depending on the observer. The inter-observer variability for the two observer groups was $2.68\mathrm{mm}\pm 1.69\mathrm{mm}$ and $3.06\mathrm{mm}\pm 1.74\mathrm{mm}$ . The observer survey and motion analysis revealed that ultrasound artifacts significantly affected labeling accuracy due to limited landmark visibility, whereas motion amplitude had no measurable effect. Our measured mean landmark motion was $11.56\mathrm{mm}\pm 5.86\mathrm{mm}$ .

Conclusions: We highlight variability in expert-labeled ground truth data for 4D ultrasound imaging and identify ultrasound artifacts as a major source of labeling inaccuracies. These findings underscore the importance of addressing observer variability and artifact-related challenges to improve the reliability of ground truth data for evaluating target tracking algorithms in 4D ultrasound applications.

Purpose: The combination of multi-layer flat panel detector (FPDT) X-ray imaging and physics-based material decomposition algorithms allows for the removal of anatomical structures. However, the reliability of these algorithms may be compromised by unaccounted materials or scattered radiation.

Approach: We investigated the two-material decomposition performance of a multi-layer FPDT in the context of 2D chest radiography without and with a 13:1 anti-scatter grid employed. A matrix-based material decomposition (MBMD) (equivalent to weighted logarithmic subtraction), a matrix-based material decomposition with polynomial beam hardening pre-correction (MBMD-PBC), and a projection domain decomposition were evaluated. The decomposition accuracy of simulated data was evaluated by comparing the bone and soft tissue images to the ground truth using the structural similarity index measure (SSIM). Simulation results were supported by experiments using a commercially available triple-layer FPDT retrofitted to a digital X-ray system.

Results: Independent of the selected decomposition algorithm, uncorrected scatter leads to negative bone estimates, resulting in small SSIM values and bone structures to remain visible in soft tissue images. Even with a 13:1 anti-scatter grid employed, bone images continue to show negative bone estimates, and bone structures appear in soft tissue images. Adipose tissue on the contrary has an almost negligible effect.

Conclusions: In a contact scan, scattered radiation leads to negative bone contrast estimates in the bone images and remaining bone contrast in the soft tissue images. Therefore, accurate scatter estimation and correction algorithms are essential when aiming for material decomposition using image data obtained with a multi-layer FPDT.

Purpose: Artificial intelligence (AI)-based medical imaging devices often include lesion or organ segmentation capabilities. Existing methods for segmentation performance evaluation compare AI results with an aggregated reference standard using accuracy metrics such as the Dice coefficient or Hausdorff distance. However, these approaches are limited by lacking a gold standard and challenges in defining meaningful success criteria. To address this, we developed a statistical method to assess agreement between an AI device and multiple human experts without requiring a reference standard.

Approach: We propose a paired-testing method to evaluate whether an AI device's segmentation performance significantly differs from that of multiple human experts. The method compares device-to-expert dissimilarity with expert-to-expert dissimilarity, avoiding the need for a reference standard. We validated the method through (1) statistical simulations where the Dice coefficient performance is either shared ("overlap agreeable") or not shared ("overlap disagreeable") between the device and experts; (2) image-based simulations using 2D contours with shared or nonshared transformation parameters (transformation agreeable or disagreeable). We also applied the method to compare an AI segmentation algorithm with four radiologists using data from the Lung Image Database Consortium.

Results: Statistical simulations show the method controls type I error ( $\sim 0.05$ ) for overlap-agreeable and type II error ( $\sim 0$ ) for overlap-disagreeable scenarios. Image-based simulations show acceptable performance with a mean type I error of 0.07 (SD 0.03) for transformation-agreeable and a mean type II error of 0.07 (SD 0.18) for transformation-disagreeable cases.

Conclusions: The paired-testing method offers a new tool for assessing the agreement between an AI segmentation device and multiple human expert panelists without requiring a reference standard.

Purpose: Many deep generative models have been proposed to reconstruct pseudo-healthy images for anomaly detection. Among these models, the variational autoencoder (VAE) has emerged as both simple and efficient. Although significant progress has been made in refining the VAE within the field of computer vision, these advancements have not been extensively applied to medical imaging applications.

Approach: We present a benchmark that assesses the ability of multiple VAEs to reconstruct pseudo-healthy neuroimages for anomaly detection in the context of dementia. We first propose a rigorous methodology to define the optimal architecture of the vanilla VAE and select through random searches the best hyperparameters of the VAE variants. Relying on a simulation-based evaluation framework, we thoroughly assess the ability of 20 VAE models to reconstruct pseudo-healthy images for the detection of dementia-related anomalies in 3D brain ${}^{18}F$ -fluorodeoxyglucose (FDG) positron emission tomography (PET) and compare their performance.

Results: This benchmark demonstrated that the majority of the VAE models tested were able to reconstruct images of good quality and generate healthy-looking images from simulated images presenting anomalies.

Conclusions: Even if no model clearly outperformed all the others, the benchmark allowed identifying a few models that perform slightly better than the vanilla VAE. It further showed that many VAE-based models can generalize to the detection of anomalies of various intensities, shapes, and locations in 3D brain FDG PET.

Purpose: State space models have shown promise in medical image segmentation by modeling long-range dependencies with linear complexity. However, they are limited in their ability to capture local features, which hinders their capacity to extract multiscale details and integrate global and local contextual information effectively. To address these shortcomings, we propose the dual-path multi-scale Mamba UNet (DMM-UNet) model.

Approach: This architecture facilitates deep fusion of local and global features through multi-scale modules within a U-shaped encoder-decoder framework. First, we introduce the multi-scale channel attention selective scanning block in the encoder, which combines global selective scanning with multi-scale channel attention to model both long-range and local dependencies simultaneously. Second, we design the spatial attention selective scanning block for the decoder. This block integrates global scanning with spatial attention mechanisms, enabling precise aggregation of semantic features through gated weighting. Finally, we develop the multi-dimensional collaborative attention layer to extract complementary attention weights across height, width, and channel dimensions, facilitating cross-space-channel feature interactions.

Results: Experiments were conducted on the ISIC17, ISIC18, Synapse, and ACDC datasets. One of the indicators, Dice similarity coefficient, achieved 89.88% on the ISIC17 dataset, 90.52% on the ISIC18 dataset, 83.07% on the Synapse dataset, and 92.60% on the ACDC dataset. There are also other indicators that perform well on this model.

Conclusions: The DMM-UNet model effectively addresses the shortcomings of state space models by enabling the integration of both local and global features, improving segmentation performance, and offering enhanced multiscale feature fusion for medical image segmentation tasks.

Purpose: Perceptual error is a significant cause of medical errors in radiology. Given the amount of information in a medical image, an image interpreter may become distracted by information unrelated to their search pattern. This may be especially challenging for novices. We aim to examine teaching medical trainees to evaluate chest radiographs (CXRs) for pulmonary nodules on limited field-of-view (LFOV) images, with the field of view (FOV) restricted to the lungs and mediastinum.

Approach: Healthcare trainees with limited exposure to interpreting images were asked to identify pulmonary nodules on CXRs, half of which contained nodules. The control and experimental groups evaluated two sets of CXRs. After the first set, the experimental group was trained to evaluate LFOV images, and both groups were again asked to assess CXRs for pulmonary nodules. Participants were given surveys after this educational session to determine their thoughts about the training and symptoms of computer vision syndrome (CVS).

Results: There was a significant improvement in performance in pulmonary nodule identification for both the experimental and control groups, but the improvement was more considerable in the experimental group ( $p\text{-}\text{value}=0.022$ ). Survey responses were uniformly positive, and each question was statistically significant (all $p\text{-}\text{values}<0.001$ ).

Conclusions: Our results show that using LFOV images may be helpful when teaching trainees specific high-yield perceptual tasks, such as nodule identification. The use of LFOV images was associated with reduced symptoms of CVS.

Purpose: Contrast-enhanced ultrasound (CEUS) is a reliable tool to diagnose focal liver lesions, which appear ambiguous in normal B-mode ultrasound. However, interpretation of the dynamic contrast sequences can be challenging, hindering the widespread application of CEUS. We investigate the use of a deep-learning-based image classifier for determining the diagnosis-relevant feature washout from CEUS acquisitions.

Approach: We introduce a data representation, which is agnostic to data heterogeneity regarding lesion size, subtype, and length of the sequences. Then, an image-based classifier is exploited for washout classification. Strategies to cope with sparse annotations and motion are systematically evaluated, as well as the potential benefits of using a perfusion model to cover missing time points.

Results: Results indicate decent performance comparable to studies found in the literature, with a maximum balanced accuracy of 84.0% on the validation and 82.0% on the test set. Correlation-based frame selection yielded improvements in classification performance, whereas further motion compensation did not show any benefit in the conducted experiments.

Conclusions: It is shown that deep-learning-based washout classification is feasible in principle. It offers a simple form of interpretability compared with benign versus malignant classifications. The concept of classifying individual features instead of the diagnosis itself could be extended to other features such as the arterial inflow behavior. The main factors distinguishing it from existing approaches are the data representation and task formulation, as well as a large dataset size with 500 liver lesions from two centers for algorithmic development and testing.

Purpose: We aim to investigate and isolate the distinctive acoustic properties generated by chemically crosslinked microbubble clusters (CCMCs) using machine learning (ML) techniques, specifically using an anomaly detection model based on autoencoders.

Approach: CCMCs were synthesized via copper-free click chemistry and subjected to acoustic analysis using a clinical transducer. Radiofrequency data were acquired, processed, and organized into training and testing datasets for the ML models. We trained an anomaly detection model with the nonclustered microbubbles (MBs) and tested the model on the CCMCs to isolate the unique acoustics. We also had a separate set of control experiments that was performed to validate the anomaly detection model.

Results: The anomaly detection model successfully identified frames exhibiting unique acoustic signatures associated with CCMCs. Frequency domain analysis further confirmed that these frames displayed higher amplitude and energy, suggesting the occurrence of potential coalescence events. The specificity of the model was validated through control experiments, in which both groups contained only individual MBs without clustering. As anticipated, no anomalies were detected in this control dataset, reinforcing the model's ability to distinguish clustered MBs from nonclustered ones.

Conclusions: We highlight the feasibility of detecting and distinguishing the unique acoustic characteristics of CCMCs, thereby improving the detectability and localization of contrast agents in ultrasound imaging. The elevated acoustic amplitudes produced by CCMCs offer potential advantages for more effective contrast agent detection, which is particularly valuable in super-resolution ultrasound imaging. Both the contrast agent and the ML-based analysis approach hold promise for a wide range of applications.

Purpose: 3D ultrasound delivers high-resolution, real-time images of soft tissues, which are essential for pain research. However, manually distinguishing various tissues for quantitative analysis is labor-intensive. We aimed to automate multilayer segmentation in 3D ultrasound volumes using minimal annotated data by developing generative reinforcement network plus (GRN+), a semi-supervised multi-model framework.

Approach: GRN+ integrates a ResNet-based generator and a U-Net segmentation model. Through a method called segmentation-guided enhancement (SGE), the generator produces new images under the guidance of the segmentation model, with its weights adjusted according to the segmentation loss gradient. To prevent gradient explosion and secure stable training, a two-stage backpropagation strategy was implemented: the first stage propagates the segmentation loss through both the generator and segmentation model, whereas the second stage concentrates on optimizing the segmentation model alone, thereby refining mask prediction using the generated images.

Results: Tested on 69 fully annotated 3D ultrasound scans from 29 subjects with six manually labeled tissue layers, GRN+ outperformed all other semi-supervised methods in terms of the Dice coefficient using only 5% labeled data, despite not using unlabeled data for unsupervised training. In addition, when applied to fully annotated datasets, GRN+ with SGE achieved a 2.16% higher Dice coefficient while incurring lower computational costs compared to other models.

Conclusions: GRN+ provides accurate tissue segmentation while reducing both computational expenses and the dependency on extensive annotations, making it an effective tool for 3D ultrasound analysis in patients with chronic lower back pain.

Purpose: Accurate iodine quantification in contrast-enhanced head CT is crucial for precise diagnosis and treatment planning. Traditional CT methods, which use energy-integrating detectors and dual-exposure techniques for material discrimination, often increase patient radiation exposure and are susceptible to motion artifacts and spectral resolution loss. Photon counting detectors (PCDs), capable of acquiring multiple energy windows in a single exposure with superior energy resolution, offer a promising alternative. However, the adoption of these technological advancements requires corresponding developments in evaluation methodologies to ensure their safe and effective implementation. One critical area of concern is the accuracy of iodine quantification, which is commonly assessed using cylindrical phantoms that neither replicate the shape of the human head nor incorporate skull-mimicking materials. These phantoms are widely used not only for testing but also for calibration, which may contribute to an overestimation of system performance in clinical applications. We address the impact of phantom design on evaluation metrics in spectral head CT, comparing conventional cylindrical phantoms to anatomically realistic elliptical phantoms with skull simulants.

Approach: We conducted simulations using a photon-counting spectral CT system equipped with cadmium telluride (CdTe) detectors, utilizing the Photon Counting Toolkit and Tigre CT software for detector response and CT geometry simulations. We compared cylindrical phantoms (20 cm diameter) to elliptical phantoms in three different sizes, incorporating skull materials with major/minor diameters and skull thicknesses of 18/14/0.5, 20/16/0.6, and 23/18/0.7 cm. Iodine inserts at concentrations of 0, 2, 5, and $10\mathrm{mg}/\mathrm{mL}$ with diameters of 1, 0.5, and 0.3 cm were used. We evaluated the influence of bowtie filters, various tube currents, and operating voltages. Image reconstruction was performed after beam hardening correction using the signal-to-thickness calibration (STC) method with standard filtered back projection, followed by both image-based and projection-based material decomposition.

Results: The results showed that image-based methods were more sensitive to phantom design, with cylindrical phantoms exhibiting enhanced performance compared with anatomically realistic designs across key metrics, including systematic error, root mean square error (RMSE), and precision. By contrast, the projection-based material decomposition method demonstrated greater consistency across different phantom designs and improved accuracy and precision. This highlights its potential for more reliable iodine quantification in complex geometries.

Conclusions: These findings underscore the critical importance of phantom design, especially the inclusion