Image Slices Research Articles

Non-contrasted computed tomography (NCCT) based chronic thromboembolic pulmonary hypertension (CTEPH) automatic diagnosis using cascaded network with multiple instance learning

Objective.The diagnosis of chronic thromboembolic pulmonary hypertension (CTEPH) is challenging due to nonspecific early symptoms, complex diagnostic processes, and small lesion sizes. This study aims to develop an automatic diagnosis method for CTEPH using non-contrasted computed tomography (NCCT) scans, enabling automated diagnosis without precise lesion annotation.Approach.A novel cascade network (CN) with multiple instance learning (CNMIL) framework was developed to improve the diagnosis of CTEPH. This method uses a CN architecture combining two Resnet-18 CNN networks to progressively distinguish between normal and CTEPH cases. Multiple instance learning (MIL) is employed to treat each 3D CT case as a 'bag' of image slices, using attention scoring to identify the most important slices. An attention module helps the model focus on diagnostically relevant regions within each slice. The dataset comprised NCCT scans from 300 subjects, including 117 males and 183 females, with an average age of 52.5 ± 20.9 years, consisting of 132 normal cases and 168 cases of lung diseases, including 88 cases of CTEPH. The CNMIL framework was evaluated using sensitivity, specificity, and the area under the curve (AUC) metrics, and compared with common 3D supervised classification networks and existing CTEPH automatic diagnosis networks.Main results. The CNMIL framework demonstrated high diagnostic performance, achieving an AUC of 0.807, accuracy of 0.833, sensitivity of 0.795, and specificity of 0.849 in distinguishing CTEPH cases. Ablation studies revealed that integrating MIL and the CN significantly enhanced performance, with the model achieving an AUC of 0.978 and perfect sensitivity (1.000) in normal classification. Comparisons with other 3D network architectures confirmed that the integrated model outperformed others, achieving the highest AUC of 0.8419.Significance. The CNMIL network requires no additional scans or annotations, relying solely on NCCT. This approach can improve timely and accurate CTEPH detection, resulting in better patient outcomes.

Cervical‐YOSA: Utilizing prompt engineering and pre‐trained large‐scale models for automated segmentation of multi‐sequence MRI images in cervical cancer

AbstractCervical cancer is a major health concern, particularly in developing countries with limited medical resources. This study introduces two models aimed at improving cervical tumor segmentation: a semi‐automatic model that fine‐tunes the Segment Anything Model (SAM) and a fully automated model designed for efficiency. Evaluations were conducted using a dataset of 8586 magnetic resonance imaging (MRI) slices, where the semi‐automatic model achieved a Dice Similarity Coefficient (DSC) of 0.9097, demonstrating high accuracy. The fully automated model also performed robustly with a DSC of 0.8526, outperforming existing methods. These models offer significant potential to enhance cervical cancer diagnosis and treatment, especially in resource‐limited settings.

Micro to macro scale anatomical analysis of the human hippocampal arteries with synchrotron hierarchical phase-contrast tomography.

To date, no non-invasive imaging modality has been employed to profile the structural intricacies of the hippocampal arterial microvasculature in humans. We hypothesised that synchrotron-based imaging of the human hippocampus would enable precise characterisation of the arterial microvasculature. Two preserved human brains from, a 69-year-old female and a 63-year-old male body donors were imaged using hierarchical phase-contrast tomography (HiP-CT) with synchrotron radiation at multiple voxel resolutions from 25.08μm down to 2.45μm. Subsequent manual and semi-automatic artery segmentation were performed followed by morphometric analyses. These data were compared to published data from alternative methodologies. HiP-CT made it possible to segment in context the arterial architecture of the human hippocampus. Our analysis identified anterior, medial and posterior hippocampal arteries arising from the P2 segment of the posterior cerebral artery on the image slices. We mapped arterial branches with external diameters greater than 50 μm in the hippocampal region. We visualised vascular asymmetry and quantified arterial structures with diameters as small as 7 μm. Through the application of HiP-CT, we have provided the first imaging visualisation and quantification of the arterial system of the human hippocampus at high resolution in the context of whole brain imaging. Our results bridge the gap between anatomical and histological scales.

Extent of Traumatic Spinal Cord Injury Is Lesion Level Dependent and Predictive of Recovery: A Multicenter Neuroimaging Study.

Assessing the extent of the intramedullary lesion after spinal cord injury (SCI) might help to improve prognostication. However, because the neurological level of injury impacts the recovery potential of SCI patients, the question arises whether lesion size parameters and predictive models based on those parameters are affected as well. In this retrospective observational study, the extent of the intramedullary lesion between individuals who sustained cervical and thoracolumbar SCI was compared, and its relation to clinical recovery was assessed. In total, 154 patients with subacute SCI (89 individuals with cervical lesions and 65 individuals with thoracolumbar lesions) underwent conventional clinical magnetic resonance imaging 1 month after injury and clinical examination at 1 and 12 months. The morphology of the focal lesion within the spinal cord was manually assessed on the midsagittal slice of T2-weighted magnetic resonance images and compared between cervical and thoracolumbar SCI patients, as well as between patients who improved at least one American Spinal Injury Association Impairment Scale (AIS) grade (converters) and patients without AIS grade improvement (nonconverters). The predictive value of lesion parameters including lesion length, lesion width, and preserved tissue bridges for predicting AIS grade conversion was assessed using regression models (conditional inference tree analysis). Lesion length was two times longer in thoracolumbar compared with cervical SCI patients (F = 39.48, p < 0.0001), whereas lesion width and tissue bridges width did not differ. When comparing AIS grade converters and nonconverters, converters showed a smaller lesion length (F = 5.46, p = 0.021), a smaller lesion width (F = 13.75, p = 0.0003), and greater tissue bridges (F = 12.87, p = 0.0005). Using regression models, tissue bridges allowed more refined subgrouping of patients in AIS groups B, C, and D according to individual recovery profiles between 1 month and 12 months after SCI, whereas lesion length added no additional information for further subgrouping. This study characterizes differences in the anteroposterior and craniocaudal lesion extents after SCI. The two times greater lesion length in thoracolumbar compared with cervical SCI might be related to differences in the anatomy, biomechanics, and perfusion between the cervical and thoracic spines. Preserved tissue bridges were less influenced by the lesion level while closely related to the clinical impairment. These results highlight the robustness and utility of tissue bridges as a neuroimaging biomarker for predicting the clinical outcome after SCI in heterogeneous patient populations and for patient stratification in clinical trials.

Intelligent void identification of particle packing system of caved ore and rock

Analyzing the void structure of the packing system of caved ore and rock and establishing the causal relationship between void structure and macroscopic properties are critical research areas in the field of gravity flow. However, the current method of analyzing void structure primarily involves manual threshold segmentation, leading to low analysis efficiency and the need for improved accuracy. This study introduces a novel approach using a dataset of two-dimensional Computed Tomography (CT) slices of an irregular limestone particle packing system. The primary goal is to enhance existing algorithms by focusing on data and network aspects to address the challenges in identifying this type of image dataset. As a result, a model named Data Attention gate with Recurrent Residual convolutional neural network based on U-net (DA2RU-net) is developed and evaluated for its convergence, accuracy, and robustness. The findings indicate the following: (1) To identify voids in the packing system of caved ore and rock using CT slices, it is advisable to employ 1000 images with dimensions between 300 and 400 pixels. It is essential to ensure that both large and small voids are represented in the image dataset. (2) The DA2RU-net model demonstrates superior performance with an average Dice value of 0.9859 for identifying large voids and 0.9654 for identifying small voids, surpassing other iterations of the U-net model and traditional algorithms. (3) The DA2RU-net model shows robustness to variations in brightness levels.

Self-normalization for a 1 mm3 resolution clinical PET system using deep learning

Objective. This work proposes, for the first time, an image-based end-to-end self-normalization framework for positron emission tomography (PET) using conditional generative adversarial networks (cGANs). Approach. We evaluated different approaches by exploring each of the following three methodologies. First, we used images that were either unnormalized or corrected for geometric factors, which encompass all time-invariant factors, as input data types. Second, we set the input tensor shape as either a single axial slice (2D) or three contiguous axial slices (2.5D). Third, we chose either Pix2Pix or polarized self-attention (PSA) Pix2Pix, which we developed for this work, as a deep learning network. The targets for all approaches were the axial slices of images normalized using the direct normalization method. We performed Monte Carlo simulations of ten voxelized phantoms with the SimSET simulation tool and produced 26,000 pairs of axial image slices for training and testing. Main results. The results showed that 2.5D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the best performance among all the methods we tested. All approaches improved general image quality figures of merit peak signal to noise ratio (PSNR) and structural similarity index (SSIM) from ∼15 % to ∼55 %, and 2.5D PSA Pix2Pix showed the highest PSNR (28.074) and SSIM (0.921). Lesion detectability, measured with region of interest (ROI) PSNR, SSIM, normalized contrast recovery coefficient, and contrast-to-noise ratio, was generally improved for all approaches, and 2.5D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the highest ROI PSNR (28.920) and SSIM (0.973). Significance. This study demonstrates the potential of an image-based end-to-end self-normalization framework using cGANs for improving PET image quality and lesion detectability without the need for separate normalization scans.

Protective effects of extracts of forsythia suspensa and chrysanthemum on alcoholic hypertension rats based on tissue pathological slice image analysis

To explore the effects of Forsythia suspensa and Chrysanthemum indicum extracts on blood pressure as well as hepatic and renal functions of hypertensive rats and to optimize the most beneficial combination of the two herbs to achieve this goal. After hypertension was observed, rats were administrated with extracts of F. suspensa and C. indicum by gastric perfusions at doses of 20 mL/kg, for fourteen days. On the fourteenth day, rats were fasted for 12 h and then blood was collected from the eyeballs for serum measurement of alanine transaminase, aspartate aminotransferase, alkaline phosphatase, albumin, total cholesterol, triglycerides, very low density lipoprotein, sodium, potassium and creatinine. Hepatic and renal tissue were harvested and processed into paraffin blocks for observation. F. suspensa and C. indicum at a ratio of 2:1 was the most effective with respect to liver repair in the hypertensive rats (p < 0.05). Both plant extracts decreased serum lipids and creatinine with the ratio of 2:1 of the two extracts being the most beneficial (p < 0.05). A synergistic effect on decreasing blood pressure and protection of the liver was observed in the hypertensive rats when extracts of F. suspensa and C. indicum were administered in combination. F. suspensa and C. indicum at a ratio of 2:1 was the most effective.

Evaluation of multiple deep neural networks for detection of intracranial dural arteriovenous fistula on susceptibility weighted angiography imaging.

The natural history of intracranial dural arteriovenous fistula (DAVF) is variable and early diagnosis is crucial in order to positively impact the clinical course of aggressive DAVF. Artificial intelligence (AI) based techniques can be promising in this regard, and in this study, we used various deep neural network (DNN) architectures to determine whether DAVF could be reliably identified on susceptibility-weighted angiography images (SWAN). A total of 3965 SWAN image slices from 30 digital subtraction angiographically proven DAVF patients and 4380 SWAN image slices from 40 age-matched patients with normal MRI findings as control group were included. The images were categorized as either DAVF or normal and the data was trained using various DNN such as VGG-16, EfficientNet-B0, and ResNet-50. Various DNN architectures showed the accuracy of 95.96% (VGG-16), 91.75% (EfficientNet-B0), and 86.23% (ResNet-50) on the SWAN image dataset. ROC analysis yielded an area under the curve of 0.796 (p < .001), best for VGG-16 model. Criterion of seven consecutive positive slices for DAVF diagnosis yielded a sensitivity of 74.68% with a specificity of 69.15%, while setting eight slices improved the sensitivity to above 80.38%, with a decrease of specificity up to 56.38%. Based on seven consecutive positive slices criteria, EfficientNet-B0 yielded a sensitivity of 73.21% with a specificity of 45.92% and ResNet-50 yielded a sensitivity of 72.39% with a specificity of 67.42%. This study shows that DNN can extract discriminative features of SWAN for the classification of DAVF from normal with good accuracy, reasonably good sensitivity and specificity.

Precise and efficient measurement of tibial slope on magnetic resonance imaging (MRI): two novel autonomous pipelines by traditional and deep learning algorithms.

The measurement of posterior tibial slopes (PTS) can aid in the screening and prevention of anterior cruciate ligament (ACL) injuries and improve the success rate of some other knee surgeries. However, the circle method for measuring PTS on magnetic resonance imaging (MRI) scans is challenging and time-consuming for most clinicians to implement in practice, despite being highly repeatable. Currently, there is no automated measurement scheme based on this method. To enhance measurement efficiency, consistency, and reduce errors resulting from manual measurements by physicians, this study proposes two novel, precise, and computationally efficient pipelines for autonomous measurement of PTS. The first pipeline employs traditional algorithms with experimental parameters to extract the tibial contour, detect adhesions, and then remove these adhesions from the extracted contour. A cyclic process is employed to adjust the parameters adaptively and generate a better binary image for the following tibial contour extraction step. The second pipeline utilizes deep learning models for classifying MRI slice images and segmenting tibial contours. The incorporation of deep learning models greatly simplifies the corresponding steps in pipeline 1. To evaluate the practical performance of the proposed pipelines, doctors utilized MRI images from 20 patients. The success rates of pipeline 1 for central, medial, and lateral slices were 85%, 100%, and 90%, respectively, while pipeline 2 achieved success rates of 100%, 100%, and 95%. Compared to the 10 minutes required for manual measurement, our automated methods enable doctors to measure PTS within 10 seconds. These evaluation results validate that the proposed pipelines are highly reliable and effective. Employing these tools can effectively prevent medical practitioners from being burdened by monotonous and repetitive manual measurement procedures, thereby enhancing both the precision and efficiency. Additionally, this tool holds the potential to contribute to the researches regarding the significance of PTS, particularly those demanding extensive and precise PTS measurement outcomes.

Three-dimensional quantitative characterization of defects in inconel 625 superalloy based on deep learning image identification

Three-dimensional (3D) quantitative characterization of defects in superalloys is an important way to promote the ability of material design and service life prediction. In this work, 3D spatial distribution of defects for Inconel 625 superalloy manufactured by laser additive manufacturing (LAM) is carried out deep learning (DL) image identification technology and 3D image reconstruction. Firstly, computer tomography (CT) technology was used to obtain continuous slice images of sample. The U-net DL algorithm was applied to intelligently identify material defects in the continuous slices. On this basis, quantitative identification and analysis of spatial defect positions and typical sizes is achieved by using 3D reconstruction software. Compared with traditional threshold segmentation (TS) techniques, the defect recognition rate has significantly improved from 61.90 ​% to 95.00 ​%. This work provides a promising characterization method for efficient characterizing alloy defects and damage especially during material performance evaluation.

Anterior interhemispheric vs. pterional approach in the microsurgical management of anterior communicating artery aneurysms: a comparative analysis employing a novel multidimensional matching-tool

The surgical management of anterior communicating artery aneurysms (AcomA) is challenging due to their deep midline position and proximity to complex skull base anatomy. This study compares the pterional craniotomy with the interhemispheric approach based on the specific aneurysm angulation. A total of 129 AcomA cases were analyzed, with 50 undergoing microsurgical clipping via either the pterional or interhemispheric approach. All selected cases had computed tomography-angiography with sagittal imaging slices and 2D-angiography. Using an interactive tool, 14 cases treated via the interhemispheric approach were matched with 14 cases approached pterionally based on clinical and morphological parameters, emphasizing intracranial aneurysm (IA) dome angulation relative to the frontal skull base. Outcomes included IA occlusion, temporary clipping incidence, intraoperative rupture, postoperative strokes, hemorrhages, hydrocephalus, vasospasm, and patient functionality. Matched cohorts had consistent demographics. Both approaches resulted in similar IA occlusion rates, but the interhemispheric approach led to improved clinical outcomes, measured by the modified Rankin Scale. It also had a lower incidence of hydrocephalus and reduced need for permanent ventriculoperitoneal shunt placement. Vasospasms and postoperative infarction rates were comparable between the groups. Our findings suggest potential advantages of the interhemispheric approach in managing AcomA, depending on aneurysm angulation. Despite a small sample size, the results highlight the importance of customized surgical decision-making based on the unique traits of each aneurysm and the surgeon's expertise.

The Use of Acoustic Mapping to Improve Surveillance of Pediatric Patient with Residual COVID-19 Symptomology

Objective: This study was designed to assess the severity of lung tissue changes due to COVID-19 in a pediatric cohort that still complained of post-COVID-19 symptoms. Materials and Methods: Parents and adolescents, 13 to 22 years old and have had a confirmatory positive polymerase chain reaction (+PCR) and/or rapid antigen test for COVID-19, consented to have a series of diagnostic tests to determine the level of lung tissue changes due to COVID-19. Sixteen adolescents completed the intake form and lung sonography. The lung sonogram was completed using the bedside lung ultrasound in emergency (BLUE) point method and a GE Logiq I laptop ultrasound system, as well as a linear 7.5-MHz or 3.0-MHz convex-array transducers. Thirteen adolescents completed a chest magnetic resonance imaging (MRI) examination on a 0.55T Siemens MAGNETOM Free.Max scanner. The examination included the heart and lungs of each participant and took about 90 minutes. A radiologist, sonographer, and computed tomography (CT) technologist reviewed and scored the images. Those positive lung sonogram scores were matched anatomically to the same slice within the chest MRI examination. Results: Based on the 13 participants, the acoustic mapping process demonstrated no diagnostic matches for positively scored sonograms at the pleural line or pulmonary parenchyma with the corresponding MRI image slices. There were diagnostic matches for negatively scored sonograms at the pleural line or pulmonary parenchyma with the corresponding MRI image slices. Conclusion: With lingering post-COVID-19 symptomology, this adolescent cohort demonstrated that acoustic mapping of their lung sonograms and chest MRI images had important combined diagnostic value. Lung sonography demonstrated diagnostic effectiveness for evaluating the periphery of the lung tissue, in this cohort but limited in viewing deeper structures. Conversely, MRI provided definitive views of the entire chest cavity and lung tissue, which is important for properly evaluating deeper pulmonary parenchyma.

Petal segmentation in CT images based on divide-and-conquer strategy.

Manual segmentation of the petals of flower computed tomography (CT) images is time-consuming and labor-intensive because the flower has many petals. In this study, we aim to obtain a three-dimensional (3D) structure of Camellia japonica flowers and propose a petal segmentation method using computer vision techniques. Petal segmentation on the slice images fails by simply applying the segmentation methods because the shape of the petals in CT images differs from that of the objects targeted by the latest instance segmentation methods. To overcome these challenges, we crop two-dimensional (2D) long rectangles from each slice image and apply the segmentation method to segment the petals on the images. Thanks to cropping, it is easier to segment the shape of the petals in the cropped images using the segmentation methods. We can also use the latest segmentation method for the task because the number of images used for training is augmented by cropping. Subsequently, the results are integrated into 3D to obtain 3D segmentation volume data. The experimental results show that the proposed method can segment petals on slice images with higher accuracy than the method without cropping. The 3D segmentation results were also obtained and visualized successfully.

Tree stumps as passive samplers for trace metal dust deposition

Atmospheric dust is contaminated by trace metals in many parts of the world due to human industrial activities. This study evaluates tree stumps as passive samplers of dust deposition and bioindicators of trace metal dust contamination in a mine impacted environment. Tree stumps (n = 230) and adjacent soils from Cu–Zn mining areas in northeastern China were analysed in situ for their trace metal concentrations using portable X-ray fluorescence (pXRF). Arsenic and Zn concentrations in tree stumps and roadside soils of each location all decreased with distance from the road in a power-law relationship, indicating a common pollution source (resuspension and redeposition of road dust). Micro-XRF images of vertical tree stump slices indicate that As and Zn were concentrated on the cut (uppermost) surface of tree stumps. This indicates atmospheric deposition as a primary contributor to elevated As and Zn concentrations in tree stumps. Trace metals in soils may have other sources besides atmospheric deposition (e.g., migration from buried ore-bodies or rock weathering), which may confound distinguishing sources in polluted areas. Here, we show that the analysis of tree stumps are suitable indicators of atmospheric trace metal contamination. Compared to surface soils, tree stumps can better reflect road dust resuspension and redeposition because unlike soils, atmospheric deposition is the primary contributor to their trace metal concentrations.

Tuned exchange imaging: Can the filter exchange imaging pulse sequence be adapted for applications with thin slices and restricted diffusion?

Filter exchange imaging (FEXI) is a double diffusion-encoding (DDE) sequence that is specifically sensitive to exchange between sites with different apparent diffusivities. FEXI uses a diffusion-encoding filtering block followed by a detection block at varying mixing times to map the exchange rate. Long mixing times enhance the sensitivity to exchange, but they pose challenges for imaging applications that require a stimulated echo sequence with crusher gradients. Thin imaging slices require strong crushers, which can introduce significant diffusion weighting and bias exchange rate estimates. Here, we treat the crushers as an additional encoding block and consider FEXI as a triple diffusion-encoding sequence. This allows the bias to be corrected in the case of multi-Gaussian diffusion, but not easily in the presence of restricted diffusion. Our approach addresses challenges in the presence of restricted diffusion and relies on the ability to independently gauge sensitivities to exchange and restricted diffusion for arbitrary gradient waveforms. It follows two principles: (i) the effects of crushers are included in the forward model using signal cumulant expansion; and (ii) timing parameters of diffusion gradients in filter and detection blocks are adjusted to maintain the same level of restriction encoding regardless of the mixing time. This results in the tuned exchange imaging (TEXI) protocol. The accuracy of exchange mapping with TEXI was assessed through Monte Carlo simulations in spheres of identical sizes and gamma-distributed sizes, and in parallel hexagonally packed cylinders. The simulations demonstrate that TEXI provides consistent exchange rates regardless of slice thickness and restriction size, even with strong crushers. However, the accuracy depends on b-values, mixing times, and restriction geometry. The constraints and limitations of TEXI are discussed, including suggestions for protocol adaptations. Further studies are needed to optimize the precision of TEXI and assess the approach experimentally in realistic, heterogeneous substrates.

A 2.5D Self-Training Strategy for Carotid Artery Segmentation in T1-Weighted Brain Magnetic Resonance Images.

Precise annotations for large medical image datasets can be time-consuming. Additionally, when dealing with volumetric regions of interest, it is typical to apply segmentation techniques on 2D slices, compromising important information for accurately segmenting 3D structures. This study presents a deep learning pipeline that simultaneously tackles both challenges. Firstly, to streamline the annotation process, we employ a semi-automatic segmentation approach using bounding boxes as masks, which is less time-consuming than pixel-level delineation. Subsequently, recursive self-training is utilized to enhance annotation quality. Finally, a 2.5D segmentation technique is adopted, wherein a slice of a volumetric image is segmented using a pseudo-RGB image. The pipeline was applied to segment the carotid artery tree in T1-weighted brain magnetic resonance images. Utilizing 42 volumetric non-contrast T1-weighted brain scans from four datasets, we delineated bounding boxes around the carotid arteries in the axial slices. Pseudo-RGB images were generated from these slices, and recursive segmentation was conducted using a Res-Unet-based neural network architecture. The model's performance was tested on a separate dataset, with ground truth annotations provided by a radiologist. After recursive training, we achieved an Intersection over Union (IoU) score of (0.68 ± 0.08) on the unseen dataset, demonstrating commendable qualitative results.

Preoperative visualization of the greater occipital nerve with magnetic resonance imaging in candidates for occipital nerve decompression for headaches

Occipital nerve decompression is effective in reducing headache symptoms in select patients with migraine and occipital neuralgia. Eligibility for surgery relies on subjective symptoms and responses to nerve blocks and Onabotulinum toxin A (Botox) injections. No validated objective method exists for detecting occipital headache pathologies. The purpose of the study is to explore the potential of high-resolution Magnetic Resolution Imaging (MRI) in identifying greater occipital nerve (GON) pathologies in chronic headache patients. The MRI protocol included three sequences targeting fat-suppressed fluid-sensitive T2-weighted signals. Visualization of the GON involved generating 2-D image slices with sequential rotation to track the nerve course. Twelve patients underwent pre-surgical MRI assessment. MRI identified four main pathologies that were validated against intra-operative examination: GON entanglement by the occipital artery, increased nerve thickness and hyperintensity suggesting inflammation compared to the non-symptomatic contralateral side, early GON branching with rejoining at a distal point, and a connection between the GON and the lesser occipital nerve. MRI possesses the ability to visualize the GON and identify suspected trigger points associated with headache symptoms. This case series highlights MRI's potential to provide objective evidence of nerve pathology. Further research is warranted to establish MRI as a gold standard for diagnosing extracranial contributors in headaches.

Relationship between pericoronary adipose tissue attenuation value and image reconstruction parameters

Rationale and objectivesTo investigate the relationship between the pericoronary adipose tissue CT mean attenuation (PCATMA) measurement and image reconstruction parameters (adaptive statistical iterative reconstruction-veo (ASIR-V) percentage, kernel, and slice thickness). Materials and methodsOne hundred and ninety-eight consecutive patients underwent CT coronary angiography at 100 kilovoltage peak (kVp) (n = 102) and 120 kVp (n = 96) were included. All scans were reconstructed by three means: 1. with 11 different ASIR-V percentages, standard kernel and 0.625 mm; 2. with soft, standard, detail, and bone kernels, 60 % ASIR-V, and 0.625 mm; 3. at 0.625 mm and 1.25 mm slice thickness, standard kernel and 60 % ASIR-V. PCATMA of the three main coronary arteries was calculated using a dedicated software. Linear regression, analysis of variance (ANOVA), Friedman test, and paired t-test were used for statistical analysis. ResultsLinear regression of pooled average data showed that the PCATMA was positively and linearly correlated with the ASIR-V percentage (all R squared >0.99). Regression analysis of individual data showed that most R squared were greater than 0.8 or 0.9, but their slope consisted of a relatively wide range. The difference of PCATMA among different kernels for each coronary artery reached statistically significant levels (P < 0.001), particularly for the difference between standard and bone kernel. Most of the differences between 0.625 mm and 1.25 mm for LAD, LCX, and RCA at 100 kVp and 120 kVp reached statistical significance (P < 0.001). ConclusionsPCATMA correlates linearly with the strength of ASIR-V. Reconstruction kernel and slice thickness also affect PCATMA, especially for the sharp kernels.

Generative adversarial network-based reconstruction of healthy anatomy for anomaly detection in brain CT scans.

To help radiologists examine the growing number of computed tomography (CT) scans, automatic anomaly detection is an ongoing focus of medical imaging research. Radiologists must analyze a CT scan by searching for any deviation from normal healthy anatomy. We propose an approach to detecting abnormalities in axial 2D CT slice images of the brain. Although much research has been done on detecting abnormalities in magnetic resonance images of the brain, there is little work on CT scans, where abnormalities are more difficult to detect due to the low image contrast that must be represented by the model used. We use a generative adversarial network (GAN) to learn normal brain anatomy in the first step and compare two approaches to image reconstruction: training an encoder in the second step and using iterative optimization during inference. Then, we analyze the differences from the original scan to detect and localize anomalies in the brain. Our approach can reconstruct healthy anatomy with good image contrast for brain CT scans. We obtain median Dice scores of 0.71 on our hemorrhage test data and 0.43 on our test set with additional tumor images from publicly available data sources. We also compare our models to a state-of-the-art autoencoder and a diffusion model and obtain qualitatively more accurate reconstructions. Without defining anomalies during training, a GAN-based network was used to learn healthy anatomy for brain CT scans. Notably, our approach is not limited to the localization of hemorrhages and tumors and could thus be used to detect structural anatomical changes and other lesions.

3D stochastic microstructure reconstruction via slice images and attention-mechanism-based GAN

Stochastic media are used to characterize materials with irregular structure and spatial randomness, and the remarkable macroscopic features of stochastic media are often determined by their internal microstructure. Hardware loads and computational burdens have always been a challenge for the reconstruction of large-volume materials. To tackle the aforementioned concerns, this paper proposes a learning model based on generative adversarial network that uses multiple 2D slice images to reconstruct 3D stochastic microstructures. The whole model training process requires only a 3D image of stochastic media as the training image. In addition, the attention mechanism captures cross-dimensional interactions to prioritize the learned features and improves the effectiveness of training. The model is tested on stochastic porous media with two-phase internal structure and complex morphology. The experimental findings demonstrate that utilizing multiple 2D images helps the model learn better and reduces the occurrence of overfitting, while greatly reducing the hardware loads of the model.