Journal of Medical Imaging最新文献

Purpose: The limited volume of medical training data remains one of the leading challenges for machine learning for diagnostic applications. Object detectors that identify and localize pathologies require training with a large volume of labeled images, which are often expensive and time-consuming to curate. To reduce this challenge, we present a method to support distant supervision of object detectors through generation of synthetic pathology-present labeled images.

Approach: Our method employs the previously proposed cyclic generative adversarial network (cycleGAN) with two key innovations: (1) use of "near-pair" pathology-present regions and pathology-absent regions from similar locations in the same subject for training and (2) the addition of a realism metric (Fréchet inception distance) to the generator loss term. We trained and tested this method with 2800 fracture-present and 2800 fracture-absent image patches from 704 unique pediatric chest radiographs. The trained model was then used to generate synthetic pathology-present images with exact knowledge of location (labels) of the pathology. These synthetic images provided an augmented training set for an object detector.

Results: In an observer study, four pediatric radiologists used a five-point Likert scale indicating the likelihood of a real fracture (1 = definitely not a fracture and 5 = definitely a fracture) to grade a set of real fracture-absent, real fracture-present, and synthetic fracture-present images. The real fracture-absent images scored $1.7\pm 1.0$, real fracture-present images $4.1\pm 1.2$, and synthetic fracture-present images $2.5\pm 1.2$. An object detector model (YOLOv5) trained on a mix of 500 real and 500 synthetic radiographs performed with a recall of $0.57\pm 0.05$ and an $F2$ score of $0.59\pm 0.05$. In comparison, when trained on only 500 real radiographs, the recall and $F2$ score were $0.49\pm 0.06$ and $0.53\pm 0.06$, respectively.

Conclusions: Our proposed method generates visually realistic pathology and that provided improved object detector performance for the task of rib fracture detection.

Purpose: We present a method that combines compressed sensing with parallel imaging that takes advantage of the structure of the sparsifying transformation.

Approach: Previous work has combined compressed sensing with parallel imaging using model-based reconstruction but without taking advantage of the structured sparsity. Blurry images for each coil are reconstructed from the fully sampled center region. The optimization problem of compressed sensing is modified to take these blurry images into account, and it is solved to estimate the missing details.

Results: Using data of brain, ankle, and shoulder anatomies, the combination of compressed sensing with structured sparsity and parallel imaging reconstructs an image with a lower relative error than does sparse SENSE or L1 ESPIRiT, which do not use structured sparsity.

Conclusions: Taking advantage of structured sparsity improves the image quality for a given amount of data as long as a fully sampled region centered on the zero frequency of the appropriate size is acquired.

Purpose: Aortic dissection carries a mortality as high as 50%, but surgical palliation is also fraught with morbidity risks of stroke or paralysis. As such, a significant focus of medical decision making is on longitudinal aortic diameters. We hypothesize that three-dimensional (3D) modeling affords a more efficient methodology toward automated longitudinal aortic measurement. The first step is to automate the measurement of manually segmented 3D models of the aorta. We developed and validated an algorithm to analyze a 3D segmented aorta and output the maximum dimension of minimum cross-sectional areas in a stepwise progression from the diaphragm to the aortic root. Accordingly, the goal is to assess the diagnostic validity of the 3D modeling measurement as a substitute for existing 2D measurements.

Approach: From January 2021 to June 2022, 66 3D non-contrast steady-state free precession magnetic resonance images of aortic pathology with clinical aortic measurements were identified; 3D aorta models were manually segmented. A novel mathematical algorithm was applied to each model to generate maximal aortic diameters from the diaphragm to the root, which were then correlated to clinical measurements.

Results: With a 76% success rate, we analyzed the resulting 50 3D aortic models utilizing the automated measurement tool. There was an excellent correlation between the automated measurement and the clinical measurement. The intra-class correlation coefficient and $p$-value for each of the nine measured locations of the aorta were as follows: sinus of valsalva, 0.99, $<0.001$; sino-tubular junction, 0.89, $<0.001$; ascending aorta, 0.97, $<0.001$; brachiocephalic artery, 0.96, $<0.001$; transverse segment 1, 0.89, $<0.001$; transverse segment 2, 0.93, $<0.001$; isthmus region, 0.92, $<0.001$; descending aorta, 0.96, $<0.001$; and aorta at diaphragm, 0.3, $<0.001$.

Conclusions: Automating diagnostic measurements that appease clinical confidence is a critical first step in a fully automated process. This tool demonstrates excellent correlation between measurements derived from manually segmented 3D models and the clinical measurements, laying the foundation for transitioning analytic methodologies from 2D to 3D.

Purpose: Accurate segmentation of the endometrium in ultrasound images is essential for gynecological diagnostics and treatment planning. Manual segmentation methods are time-consuming and subjective, prompting the exploration of automated solutions. We introduce "segment anything with inception module" (SAIM), a specialized adaptation of the segment anything model, tailored specifically for the segmentation of endometrium structures in ultrasound images.

Approach: SAIM incorporates enhancements to the image encoder structure and integrates point prompts to guide the segmentation process. We utilized ultrasound images from patients undergoing hysteroscopic surgery in the gynecological department to train and evaluate the model.

Results: Our study demonstrates SAIM's superior segmentation performance through quantitative and qualitative evaluations, surpassing existing automated methods. SAIM achieves a dice similarity coefficient of 76.31% and an intersection over union score of 63.71%, outperforming traditional task-specific deep learning models and other SAM-based foundation models.

Conclusions: The proposed SAIM achieves high segmentation accuracy, providing high diagnostic precision and efficiency. Furthermore, it is potentially an efficient tool for junior medical professionals in education and diagnosis.

Purpose: Statistical shape and intensity models (SSIMs) and digitally reconstructed radiographs (DRRs) were introduced for non-rigid 2D-3D registration and skeletal geometry/density reconstruction studies. The computation of DRRs takes most of the time during registration or reconstruction. The goal of this study is to propose a particle-based method for composing an SSIM and a DRR image generation scheme and analyze the quality of the images compared with previous DRR generation methods.

Approach: Particle-based SSIMs consist of densely scattered particles on the surface and inside of an object, with each particle having an intensity value. Generating the DRR resembles ray tracing, which counts the particles that are binned with each ray and calculates the radiation attenuation. The distance between adjacent particles was considered to be the radiologic path during attenuation integration, and the mean linear attenuation coefficient of the two particles was multiplied. The proposed method was compared with the DRR of CT projection. The mean squared error and peak signal-to-noise ratio (PSNR) were calculated between the DRR images from the proposed method and those of existing methods of projecting tetrahedral-based SSIMs or computed tomography (CT) images to verify the accuracy of the proposed scheme.

Results: The suggested method was about 600 times faster than the tetrahedral-based SSIM without using the hardware acceleration technique. The PSNR was 37.59 dB, and the root mean squared error of the normalized pixel intensities was 0.0136.

Conclusions: The proposed SSIM and DRR generation procedure showed high temporal performance while maintaining image quality, and particle-based SSIM is a feasible form for representing a 3D volume and generating the DRR images.

Purpose: We aim to determine the combination of X-ray spectrum and detector scintillator thickness that maximizes the detectability of microcalcification clusters in dedicated cone-beam breast CT.

Approach: A cascaded linear system analysis was implemented in the spatial frequency domain and was used to determine the detectability index using numerical observers for the imaging task of detecting a microcalcification cluster with 0.17 mm diameter calcium carbonate spheres. The analysis considered a thallium-doped cesium iodide scintillator coupled to a complementary metal-oxide semiconductor detector and an analytical filtered-back-projection reconstruction algorithm. Independent system parameters considered were the scintillator thickness, applied X-ray tube voltage, and X-ray beam filtration. The combination of these parameters that maximized the detectability index was considered optimal.

Results: Prewhitening, nonprewhitening, and nonprewhitening with eye filter numerical observers indicate that the combination of 0.525 to 0.6 mm thick scintillator, 70 kV, and 0.25 to 0.4 mm added copper filtration maximized the detectability index at a mean glandular dose (MGD) of 4.5 mGy.

Conclusion: Using parallel cascade systems' analysis, the combination of parameters that could maximize the detection of microcalcifications was identified. The analysis indicates that a harder beam than that used in current practice may be beneficial for the task of detecting microcalcifications at an MGD suitable for breast cancer screening.

Purpose: Immune checkpoint inhibitors (ICIs) are now one of the standards of care for patients with lung cancer and have greatly improved both progression-free and overall survival, although $<20\%$ of the patients respond to the treatment, and some face acute adverse events. Although a few predictive biomarkers have integrated the clinical workflow, they require additional modalities on top of whole-slide images and lack efficiency or robustness. In this work, we propose a biomarker of immunotherapy outcome derived solely from the analysis of histology slides.

Approach: We develop a three-step framework, combining contrastive learning and nonparametric clustering to distinguish tissue patterns within the slides, before exploiting the adjacencies of previously defined regions to derive features and train a proportional hazards model for survival analysis. We test our approach on an in-house dataset of 193 patients from 5 medical centers and compare it with the gold standard tumor proportion score (TPS) biomarker.

Results: On a fivefold cross-validation (CV) of the entire dataset, the whole-slide image-based survival analysis for patients treated with immunotherapy (WhARIO) features are able to separate a low- and a high-risk group of patients with a hazard ratio (HR) of 2.29 (${\mathrm{CI}}_{95}=1.48$ to 3.56), whereas the TPS 1% reference threshold only reaches a HR of 1.81 (${\mathrm{CI}}_{95}=1.21$ to 2.69). Combining the two yields a higher HR of 2.60 (${\mathrm{CI}}_{95}=1.72$ to 3.94). Additional experiments on the same dataset, where one out of five centers is excluded from the CV and used as a test set, confirm these trends.

Conclusions: Our uniquely designed WhARIO features are an efficient predictor of survival for lung cancer patients who received ICI treatment. We achieve similar performance to the current gold standard biomarker, without the need to access other imaging modalities, and show that both can be used together to reach even better results.

Purpose: The purpose of this study is to compare interpretation efficiency of radiologists reading radiographs on 6 megapixel (MP) versus 12 MP monitors.

Approach: Our method compares two sets of monitors in two phases: in phase I, radiologists interpreted using a 6 MP, 30.4 in. (Barco Coronis Fusion) and in phase II, a 12 MP, 30.9 in. (Barco Nio Fusion). Nine chest and three musculoskeletal radiologists each batch interpreted an average of 115 radiographs in phase I and 115 radiographs in phase II as a part of routine clinical work. Radiologists were blinded to monitor resolution.

Results: Interpretation times per radiograph were noted from dictation logs. Interpretation time was significantly decreased utilizing a 12 MP monitor by 6.88 s ( $p=0.002$ ) and 6.76 s (8.7%) ( $p<0.001$ ) for chest radiographs only and combined chest and musculoskeletal radiographs, respectively. When evaluating musculoskeletal radiographs alone, the improvement in reading times with 12 MP monitor was 6.76 s, however, this difference was not statistically significant ( $p=0.111$ ). Interpretation of radiographs on 12 MP monitors was 8.7% faster than on 6 MP monitors.

Conclusion: Higher resolution diagnostic displays can enable radiologists to interpret radiographs more efficiently.

Purpose: The objective of this study is to evaluate the accuracy of an augmented reality (AR) system in improving guidance, accuracy, and visualization during the subxiphoidal approach for epicardial ablation.

Approach: An AR application was developed to project real-time needle trajectories and patient-specific 3D organs using the Hololens 2. Additionally, needle tracking was implemented to offer real-time feedback to the operator, facilitating needle navigation. The AR application was evaluated through three different experiments: examining overlay accuracy, assessing puncture accuracy, and performing pre-clinical evaluations on a phantom.

Results: The results of the overlay accuracy assessment for the AR system yielded $2.36\pm 2.04\mathrm{mm}$. Additionally, the puncture accuracy utilizing the AR system yielded $1.02\pm 2.41\mathrm{mm}$. During the pre-clinical evaluation on the phantom, needle puncture with AR guidance showed $7.43\pm 2.73\mathrm{mm}$, whereas needle puncture without AR guidance showed $22.62\pm 9.37\mathrm{mm}$.

Conclusions: Overall, the AR platform has the potential to enhance the accuracy of percutaneous epicardial access for mapping and ablation of cardiac arrhythmias, thereby reducing complications and improving patient outcomes. The significance of this study lies in the potential of AR guidance to enhance the accuracy and safety of percutaneous epicardial access.

Purpose: The accurate detection and tracking of devices, such as guiding catheters in live X-ray image acquisitions, are essential prerequisites for endovascular cardiac interventions. This information is leveraged for procedural guidance, e.g., directing stent placements. To ensure procedural safety and efficacy, there is a need for high robustness/no failures during tracking. To achieve this, one needs to efficiently tackle challenges, such as device obscuration by the contrast agent or other external devices or wires and changes in the field-of-view or acquisition angle, as well as the continuous movement due to cardiac and respiratory motion.

Approach: To overcome the aforementioned challenges, we propose an approach to learn spatio-temporal features from a very large data cohort of over 16 million interventional X-ray frames using self-supervision for image sequence data. Our approach is based on a masked image modeling technique that leverages frame interpolation-based reconstruction to learn fine inter-frame temporal correspondences. The features encoded in the resulting model are fine-tuned downstream in a light-weight model.

Results: Our approach achieves state-of-the-art performance, in particular for robustness, compared to ultra optimized reference solutions (that use multi-stage feature fusion or multi-task and flow regularization). The experiments show that our method achieves a 66.31% reduction in the maximum tracking error against the reference solutions (23.20% when flow regularization is used), achieving a success score of 97.95% at a $3\times$ faster inference speed of 42 frames-per-second (on GPU). In addition, we achieve a 20% reduction in the standard deviation of errors, which indicates a much more stable tracking performance.

Conclusions: The proposed data-driven approach achieves superior performance, particularly in robustness and speed compared with the frequently used multi-modular approaches for device tracking. The results encourage the use of our approach in various other tasks within interventional image analytics that require effective understanding of spatio-temporal semantics.