RESUMO
OBJECTIVE: This study evaluated the clinical utility of a phantom-based convolutional neural network noise reduction framework for whole-body-low-dose CT skeletal surveys. MATERIALS AND METHODS: The CT exams of ten patients with multiple myeloma were retrospectively analyzed. Exams were acquired with routine whole-body-low-dose CT protocol and projection noise insertion was used to simulate 25% dose exams. Images were reconstructed with either iterative reconstruction or filtered back projection with convolutional neural network post-processing. Diagnostic quality and structure visualization were blindly rated (subjective scale ranging from 0 [poor] to 100 [excellent]) by three musculoskeletal radiologists for iterative reconstruction and convolutional neural network images at routine whole-body-low-dose and 25% dose CT. RESULTS: For the diagnostic quality rating, the convolutional neural network outscored iterative reconstruction at routine whole-body-low-dose CT (convolutional neural network: 95 ± 5, iterative reconstruction: 85 ± 8) and at the 25% dose level (convolutional neural network: 79 ± 10, iterative reconstruction: 22 ± 13). Convolutional neural network applied to 25% dose was rated inferior to iterative reconstruction applied to routine dose. Similar trends were observed in rating experiments focusing on structure visualization. CONCLUSION: Results indicate that the phantom-based convolutional neural network noise reduction framework can improve visualization of critical structures within CT skeletal surveys. At matched dose level, the convolutional neural network outscored iterative reconstruction for all conditions studied. The image quality improvement of convolutional neural network applied to 25% dose indicates a potential for dose reduction; however, the 75% dose reduction condition studied is not currently recommended for clinical implementation.
Assuntos
Redes Neurais de Computação , Tomografia Computadorizada por Raios X , Algoritmos , Humanos , Imagens de Fantasmas , Doses de Radiação , Interpretação de Imagem Radiográfica Assistida por Computador , Estudos RetrospectivosRESUMO
Iterative reconstruction (IR) algorithms are the most widely used CT noise-reduction method to improve image quality and have greatly facilitated radiation dose reduction within the radiology community. Various IR methods have different strengths and limitations. Because IR algorithms are typically nonlinear, they can modify spatial resolution and image noise texture in different regions of the CT image; hence traditional image-quality metrics are not appropriate to assess the ability of IR to preserve diagnostic accuracy, especially for low-contrast diagnostic tasks. In this review, the authors highlight emerging IR algorithms and CT noise-reduction techniques and summarize how these techniques can be evaluated to help determine the appropriate radiation dose levels for different diagnostic tasks in CT. In addition to advanced IR techniques, we describe novel CT noise-reduction methods based on convolutional neural networks (CNNs). CNN-based noise-reduction techniques may offer the ability to reduce image noise while maintaining high levels of image detail but may have unique drawbacks. Other novel CT noise-reduction methods are being developed to leverage spatial and/or spectral redundancy in multiphase or multienergy CT. Radiologists and medical physicists should be familiar with these different alternatives to adapt available CT technology for different diagnostic tasks. The scope of this article is (a) to review the clinical applications of IR algorithms as well as their strengths, weaknesses, and methods of assessment and (b) to explore new CT image reconstruction and noise-reduction techniques that promise to facilitate radiation dose reduction. ©RSNA, 2021.
Assuntos
Algoritmos , Tomografia Computadorizada por Raios X , Humanos , Processamento de Imagem Assistida por Computador , Imagens de Fantasmas , Doses de Radiação , Interpretação de Imagem Radiográfica Assistida por ComputadorRESUMO
OBJECTIVE: The aim of this study was to evaluate a narrowly trained convolutional neural network (CNN) denoising algorithm when applied to images reconstructed differently than training data set. METHODS: A residual CNN was trained using 10 noise inserted examinations. Training images were reconstructed with 275 mm of field of view (FOV), medium smooth kernel (D30), and 3 mm of thickness. Six examinations were reserved for testing; these were reconstructed with 100 to 450 mm of FOV, smooth to sharp kernels, and 1 to 5 mm of thickness. RESULTS: When test and training reconstruction settings were not matched, there was either reduced denoising efficiency or resolution degradation. Denoising efficiency was reduced when FOV was decreased or a smoother kernel was used. Resolution loss occurred when the network was applied to an increased FOV, sharper kernel, or decreased image thickness. CONCLUSIONS: The CNN denoising performance was degraded with variations in FOV, kernel, or decreased thickness. Denoising performance was not affected by increased thickness.
Assuntos
Processamento de Imagem Assistida por Computador/métodos , Redes Neurais de Computação , Razão Sinal-Ruído , Tomografia Computadorizada por Raios X/métodos , Algoritmos , Aprendizado Profundo , HumanosRESUMO
Machine learning and artificial intelligence (AI) algorithms hold significant promise for addressing important clinical needs when applied to medical imaging; however, integration of algorithms into a radiology department is challenging. Vended algorithms are integrated into the workflow, successfully, but are typically closed systems and unavailable for site researchers to deploy algorithms. Rather than AI researchers creating one-off solutions, a general, multi-purpose integration system is desired. Here, we present a set of use cases and requirements for a system designed to enable rapid deployment of AI algorithms into the radiologist's workflow. The system uses standards-compliant digital imaging and communications in medicine structured reporting (DICOM SR) to present AI measurements, results, and findings to the radiologist in a clinical context and enables acceptance or rejection of results. The system also implements a feedback mechanism for post-processing technologists to correct results as directed by the radiologist. We demonstrate integration of a body composition algorithm and an algorithm for determining total kidney volume for patients with polycystic kidney disease.
Assuntos
Inteligência Artificial , Radiologia , Algoritmos , Humanos , Radiologistas , Fluxo de TrabalhoRESUMO
BACKGROUND: Computed tomography (CT) is routinely used to guide cryoablation procedures. Notably, CT-guidance provides 3D localization of cryoprobes and can be used to delineate frozen tissue during ablation. However, metal-induced artifacts from ablation probes can make accurate probe placement challenging and degrade the ice ball conspicuity, which in combination could lead to undertreatment of potentially curable lesions. PURPOSE: In this work, we propose an image-based neural network (CNN) model for metal artifact reduction for CT-guided interventional procedures. METHODS: An image domain metal artifact simulation framework was developed and validated for deep-learning-based metal artifact reduction for interventional oncology (MARIO). CT scans were acquired for 19 different cryoablation probe configurations. The probe configurations varied in the number of probes and the relative orientations. A combination of intensity thresholding and masking based on maximum intensity projections (MIPs) was used to segment both the probes only and probes + artifact in each phantom image. Each of the probe and probe + artifact images were then inserted into 19 unique patient exams, in the image domain, to simulate metal artifact appearance for CT-guided interventional oncology procedures. The resulting 361 pairs of simulated image volumes were partitioned into disjoint training and test datasets of 304 and 57 volumes, respectively. From the training partition, 116 600 image patches with a shape of 128 × 128 × 5 pixels were randomly extracted to be used for training data. The input images consisted of a superposition of the patient and probe + artifact images. The target images consisted of a superposition of the patient and probe only images. This dataset was used to optimize a U-Net type model. The trained model was then applied to 50 independent, previously unseen CT images obtained during renal cryoablations. Three board-certified radiologists with experience in CT-guided ablations performed a blinded review of the MARIO images. A total of 100 images (50 original, 50 MARIO processed) were assessed across different aspects of image quality on a 4-point likert-type item. Statistical analyses were performed using Wilcoxon signed-rank test for paired samples. RESULTS: Reader scores were significantly higher for MARIO processed images compared to the original images across all metrics (all p < 0.001). The average scores of the overall image quality, iceball conspicuity, overall metal artifact, needle tip visualization, target region confidence, and worst metal artifact, needle tip visualization, iceball conspicuity, and target region confidence improved by 34.91%, 36.29%, 39.94%, 34.17%, 35.13%, and 45.70%, respectively. CONCLUSIONS: The proposed method of image-based metal artifact simulation can be used to train a MARIO algorithm to effectively reduce probe-related metal artifacts in CT-guided cryoablation procedures.
Assuntos
Artefatos , Processamento de Imagem Assistida por Computador , Metais , Tomografia Computadorizada por Raios X , Processamento de Imagem Assistida por Computador/métodos , Redes Neurais de Computação , Humanos , Cirurgia Assistida por Computador/métodos , Aprendizado Profundo , Criocirurgia/métodos , Imagens de FantasmasRESUMO
Coronary computed tomography angiography (cCTA) is a widely used non-invasive diagnostic exam for patients with coronary artery disease (CAD). However, most clinical CT scanners are limited in spatial resolution from use of energy-integrating detectors (EIDs). Radiological evaluation of CAD is challenging, as coronary arteries are small (3-4 mm diameter) and calcifications within them are highly attenuating, leading to blooming artifacts. As such, this is a task well suited for high spatial resolution. Recently, photon-counting-detector (PCD) CT became commercially available, allowing for ultra-high resolution (UHR) data acquisition. However, PCD-CTs are costly, restricting widespread accessibility. To address this problem, we propose a super resolution convolutional neural network (CNN): ILUMENATE (Improved LUMEN visualization through Artificial super-resoluTion imagEs), creating a high resolution (HR) image simulating UHR PCD-CT. The network was trained and validated using patches extracted from 8 patients with a modified U-Net architecture. Training input and labels consisted of UHR PCD-CT images reconstructed with a smooth kernel degrading resolution (LR input) and sharp kernel (HR label). The network learned the resolution difference and was tested on 5 unseen LR patients. We evaluated network performance quantitatively and qualitatively through visual inspection, line profiles to assess spatial resolution improvements, ROIs for CT number stability and noise assessment, structural similarity index (SSIM), and percent diameter luminal stenosis. Overall, ILUMENATE improved images quantitatively and qualitatively, creating sharper edges more closely resembling reconstructed HR reference images, maintained stable CT numbers with less than 4% difference, reduced noise by 28%, maintained structural similarity (average SSIM = 0.70), and reduced percent diameter stenosis with respect to input images. ILUMENATE demonstrates potential impact for CAD patient management, improving the quality of LR CT images bringing them closer to UHR PCD-CT images.
RESUMO
In CT imaging of the head, multiple image series are routinely reconstructed with different kernels and slice thicknesses. Reviewing the redundant information is an inefficient process for radiologists. We address this issue with a convolutional neural network (CNN)-based technique, synthesiZed Improved Resolution and Concurrent nOise reductioN (ZIRCON), that creates a single, thin, low-noise series that combines the favorable features from smooth and sharp head kernels. ZIRCON uses a CNN model with an autoencoder U-Net architecture that accepts two input channels (smooth- and sharp-kernel CT images) and combines their salient features to produce a single CT image. Image quality requirements are built into a task-based loss function with a smooth and sharp loss terms specific to anatomical regions. The model is trained using supervised learning with paired routine-dose clinical non-contrast head CT images as training targets and simulated low-dose (25%) images as training inputs. One hundred unique de-identified clinical exams were used for training, ten for validation, and ten for testing. Visual comparisons and contrast measurements of ZIRCON revealed that thinner slices and the smooth-kernel loss function improved gray-white matter contrast. Combined with lower noise, this increased visibility of small soft-tissue features that would be otherwise impaired by partial volume averaging or noise. Line profile analysis showed that ZIRCON images largely retained sharpness compared to the sharp-kernel input images. ZIRCON combined desirable image quality properties of both smooth and sharp input kernels into a single, thin, low-noise series suitable for both brain and skull imaging.
RESUMO
Automated segmentation tools often encounter accuracy and adaptability issues when applied to images of different pathology. The purpose of this study is to explore the feasibility of building a workflow to efficiently route images to specifically trained segmentation models. By implementing a deep learning classifier to automatically classify the images and route them to appropriate segmentation models, we hope that our workflow can segment the images with different pathology accurately. The data we used in this study are 350 CT images from patients affected by polycystic liver disease and 350 CT images from patients presenting with liver metastases from colorectal cancer. All images had the liver manually segmented by trained imaging analysts. Our proposed adaptive segmentation workflow achieved a statistically significant improvement for the task of total liver segmentation compared to the generic single-segmentation model (non-parametric Wilcoxon signed rank test, n = 100, p-value << 0.001). This approach is applicable in a wide range of scenarios and should prove useful in clinical implementations of segmentation pipelines.
RESUMO
Purpose: Convolutional neural networks (CNNs) have been proposed for super-resolution in CT, but training of CNNs requires high-resolution reference data. Higher spatial resolution can also be achieved using deconvolution, but conventional deconvolution approaches amplify noise. We develop a CNN that mitigates increasing noise and that does not require higher-resolution reference images. Methods: Our model includes a noise reduction CNN and a deconvolution CNN that are separately trained. The noise reduction CNN is a U-Net, similar to other noise reduction CNNs found in the literature. The deconvolution CNN uses an autoencoder, where the decoder is fixed and provided as a hyperparameter that represents the system point spread function. The encoder is trained to provide a deconvolution that does not amplify noise. Ringing can occur from deconvolution but is controlled with a difference of gradients loss function term. Our technique was demonstrated on a variety of patient images and on ex vivo kidney stones. Results: The noise reduction and deconvolution CNNs produced visually sharper images at low noise. In ex vivo mixed kidney stones, better visual delineation of the kidney stone components could be seen. Conclusions: A noise reduction and deconvolution CNN improves spatial resolution and reduces noise without requiring higher-resolution reference images.
RESUMO
BACKGROUND: Deep artificial neural networks such as convolutional neural networks (CNNs) have been shown to be effective models for reducing noise in CT images while preserving anatomic details. A practical bottleneck for developing CNN-based denoising models is the procurement of training data consisting of paired examples of high-noise and low-noise CT images. Obtaining these paired data are not practical in a clinical setting where the raw projection data is not available. This work outlines a technique to optimize CNN denoising models using methods that are available in a routine clinical setting. PURPOSE: To demonstrate a phantom-based training framework for CNN noise reduction that can be efficiently implemented on any CT scanner. METHODS: The phantom-based training framework uses supervised learning in which training data are synthesized using an image-based noise insertion technique. Ten patient image series were used for training and validation (9:1) and noise-only images obtained from anthropomorphic phantom scans. Phantom noise-only images were superimposed on patient images to imitate low-dose CT images for use in training. A modified U-Net architecture was used with mean-squared-error and feature reconstruction loss. The training framework was tested for clinically indicated whole-body-low-dose CT images, as well as routine abdomen-pelvis exams for which projection data was unavailable. Performance was assessed based on root-mean-square error, structural similarity, line profiles, and visual assessment. RESULTS: When the CNN was tested on five reserved quarter-dose whole-body-low-dose CT images, noise was reduced by 75%, root-mean-square-error reduced by 34%, and structural similarity increased by 60%. Visual analysis and line profiles indicated that the method significantly reduced noise while maintaining spatial resolution of anatomic features. CONCLUSION: The proposed phantom-based training framework demonstrated strong noise reduction while preserving spatial detail. Because this method is based within the image domain, it can be easily implemented without access to projection data.
Assuntos
Processamento de Imagem Assistida por Computador , Redes Neurais de Computação , Humanos , Processamento de Imagem Assistida por Computador/métodos , Tomografia Computadorizada por Raios X/métodos , Tomógrafos Computadorizados , Imagens de Fantasmas , Razão Sinal-RuídoRESUMO
BACKGROUND: Focal-occult placenta accreta spectrum is known to cause adverse obstetrical morbidity outcomes, however, direct comparisons with previa-associated placenta accreta spectrum morbidity are lacking. OBJECTIVE: We sought to compare the baseline characteristics, surgical and obstetrical morbidity, and subsequent pregnancy outcomes of patients with focal-occult placenta accreta spectrum with those of patients with previa-associated accreta. STUDY DESIGN: A retrospective review was conducted of all pathologically confirmed placenta accreta spectrum cases from 2018 to 2022 at a tertiary care center. The baseline characteristics, surgical, obstetrical, and subsequent pregnancy outcomes were recorded. Cases of focal-occult placenta accreta spectrum was compared with cases of previa-associated placenta accreta spectrum across a range of morbidity characteristics including hemorrhagic factors, interventions, postdelivery reoperations, infections, and intensive care unit admission. Statistical comparison was performed using Kruskal-Wallis or chi-square tests, and a P value of <.05 was considered significant. RESULTS: A total of 74 cases were identified with 43 focal-occult and 31 previa-associated placenta accreta spectrum cases. Of those, 25.6% of the patients with focal-occult placenta accreta spectrum and 100% of the patients with previa-associated placenta accreta spectrum underwent a hysterectomy. One case of focal-occult placenta accreta spectrum and 29 cases of previa-associated placenta accreta spectrum were diagnosed antenatally. Patients with focal-occult placenta accreta spectrum did not differ from those with previa-associated placenta accreta spectrum in mean maternal age (33.0 vs 33.1 years), body mass index, or the incidence of previous dilation and curettage procedures (16.3% vs 25.8%). Patients with focal-occult placenta accreta spectrum were significantly more likely to have a lower mean parity (1.5 vs 3.6 gestations), higher gestational age at delivery (36.1 vs 33.9 weeks' gestation), and were less likely to have had a previous cesarean delivery (12/43, 27.9% vs 30/31, 96.8%). In addition, patients with focal-occult placenta accreta spectrum had less previous cesarean deliveries (mean, 0.5 vs 2.3), were more likely to have undergone in vitro fertilization (20.9% vs 3.2%), and less likely to have anterior placentation. When contrasting the clinical outcomes of patients with focal-occult placenta accreta spectrum with those with previa-associated placenta accreta spectrum, the postpartum hemorrhage rates (71.0% vs 67.4%), mean quantitative blood loss (2099 mL; range, 500-9516 mL vs 2119 mL; range 350-12,220 mL), mean units of red blood cells transfused (1.4 vs 1.7), massive transfusion rate (9.3% vs 3.2%), and intensive care unit admission rates (11.6% vs 6.5%) were not significantly different, but there was a nonsignificant trend toward higher morbidity among patients with focal-occult accreta. Patients with focal-occult accreta had a higher incidence of reoperations or return to the operating room (30.2 vs 6.5%; P=.01). When comparing focal-occult with previa-associated placenta accreta spectrum, the composite outcomes, including hemorrhagic morbidity (77.4% vs 74.4%), any maternal morbidity (83.9% vs 76.7%), and severe maternal morbidity (64.5% vs 65.1%), were not significantly different between the groups. Nine focal-occult placenta accreta spectrum patients had a subsequent pregnancy, and 3 of those had recurrent placenta accreta spectrum. CONCLUSION: Focal-occult placenta accreta spectrum presents with fewer identifiable risk factors than placenta previa-associated placenta accreta spectrum but may be associated with an in vitro fertilization pregnancy. Patients with focal-occult placenta accreta spectrum was observed to have a higher incidence of reoperation when compared with patients previa-associated placenta accreta spectrum, and no other statistically significant differences in morbidity outcomes were observed. The absence of differences in morbidity outcomes may be attributable to a lack of antenatal detection of focal-occult accreta and merits further investigation.
Assuntos
Placenta Acreta , Placenta Prévia , Gravidez , Humanos , Feminino , Adulto , Lactente , Cesárea/efeitos adversos , Placenta Acreta/diagnóstico , Placenta Acreta/epidemiologia , Placenta Acreta/cirurgia , Estudos Retrospectivos , Histerectomia/métodos , Resultado da Gravidez , Placenta Prévia/diagnóstico , Placenta Prévia/epidemiologia , Placenta Prévia/etiologiaRESUMO
In this study, we describe a systematic approach to optimize deep-learning-based image processing algorithms using random search. The optimization technique is demonstrated on a phantom-based noise reduction training framework; however, the techniques described can be applied generally for other deep learning image processing applications. The parameter space explored included number of convolutional layers, number of filters, kernel size, loss function, and network architecture (either U-Net or ResNet). A total of 100 network models were examined (50 random search, 50 ablation experiments). Following the random search, ablation experiments resulted in a very minor performance improvement indicating near optimal settings were found during the random search. The top performing network architecture was a U-Net with 4 pooling layers, 64 filters, 3×3 kernel size, ELU activation, and a weighted feature reconstruction loss (0.2×VGG + 0.8×MSE). Relative to the low-dose input image, the CNN reduced noise by 90%, reduced RMSE by 34%, and increased SSIM by 76% on six patient exams reserved for testing. The visualization of hepatic and bone lesions was greatly improved following noise reduction.
RESUMO
This article shows how to train a convolutional neural network to reduce noise in CT images, although the principles apply to medical and nonmedical images; authors also explore mathematical and visually weighted loss functions to adjust the appearance.
RESUMO
PURPOSE: Filtering measured projections with a particular convolutional kernel is an essential step in analytic reconstruction of computed tomography (CT) images. A tradeoff between noise and spatial resolution exists for different choices of reconstruction kernel. In a clinical setting, this often requires producing multiple images reconstructed with different kernels for a single CT exam, which increases the burden of computation, networking, archival, and reading. We address this problem by training a deep convolutional neural network (CNN) to synthesize multiple input images into a single output image which exhibits low noise while also preserving features in images reconstructed with the sharpest kernels. METHODS: A CNN architecture consisting of repeated blocks of residual units containing a total of 20 convolutional layers was used to combine features. The CNN inputs consisted of two images produced with different reconstruction kernels, one smooth and one sharp, which were stacked in the channel dimension. The network was trained using supervised learning with both full-dose and simulated quarter-dose abdominal CT images. After training, the performance was evaluated using a reserved set of full-dose scans that were not used for network optimization. Noise reduction performance was measured by comparing root mean square (RMS) measurements in uniform regions. Spatial resolution was compared using line profiles of anatomic features. RESULTS: For the regions tested, the synthetic images feature noise levels slightly below those of the smooth input images, while maintaining the resolution of anatomic details found in the sharp input images. CONCLUSIONS: A deep CNN can be used combine features from CT images reconstructed with different kernels to produce a single synthesized image series that exhibits both low noise and high spatial resolution. This approach has implications for improving image quality, reducing radiation dose, and simplifying the clinical workflow for CT imaging.