Thoracic CT Images Research Articles

Artificial intelligence-based vessel suppression for detection of sub-solid nodules in lung cancer screening computed tomography.

Lung cancer screening (LCS) with low-dose computed tomography (LDCT) helps early lung cancer detection, commonly presenting as small pulmonary nodules. Artificial intelligence (AI)-based vessel suppression (AI-VS) and automatic detection (AI-AD) algorithm can improve detection of subsolid nodules (SSNs) on LDCT. We assessed the impact of AI-VS and AI-AD in detection and classification of SSNs [ground-glass nodules (GGNs) and part-solid nodules (PSNs)], on LDCT performed for LCS. Following regulatory approval, 123 LDCT examinations with sub-solid pulmonary nodules (average diameter ≥6 mm) were processed to generate three image series for each examination-unprocessed, AI-VS, and AI-AD series with annotated lung nodules. Two thoracic radiologists in consensus formed the standard of reference (SOR) for this study. Two other thoracic radiologists (R1 and R2; 5 and 10 years of experience in thoracic CT image interpretation) independently assessed the unprocessed images alone, then together with AI-VS series, and finally with AI-AD for detecting all ≥6 mm GGN and PSN. We performed receiver operator characteristics (ROC) and Cohen's Kappa analyses for statistical analyses. On unprocessed images, R1 and R2 detected 232/310 nodules (R1: 114 GGN, 118 PSN) and 255/310 nodules (R2: 122 GGN, 133 PSN), respectively (P>0.05). On AI-VS images, they detected 249/310 nodules (119 GGN, 130 PSN) and 277/310 nodules (128 GGN, 149 PSN), respectively (P≥0.12). When compared to the SOR, accuracy (AUC) for detection of PSN on the AI-VS images (AUC 0.80-0.81) was greater than on the unprocessed images (AUC 0.70-0.76). AI-VS images enabled detection of solid components in five nodules deemed as GGN on the unprocessed images. Accuracy of AI-AD was lower than both the radiologists (AUC 0.60-0.72). AI-VS improved the detection and classification of SSN into GGN and PSN on LDCT of the chest for the two radiologist (R1 and R2) readers.

More Than Just Suckdown: Hemodynamics of Intermittent Inflow Cannula Obstruction

<h3>Introduction</h3> Waveforms can provide important insights on patient hemodynamics after durable left ventricular assist device (LVAD) implantation. Herein we describe a case of LVAD implanted as BTC complicated by intermittent inflow cannula obstruction. <h3>Case Report</h3> A 55-year old male was admitted decompensated heart failure. He had a Heartware device implanted as a bridge to candidacy (BTC) due to pulmonary hypertension and ischemic cardiomyopathy. In the 8 months since LVAD implant, the patient had gained 25 kg weight and had struggled to maintain a euvolemic state. Examination revealed an elevated jugular venous pressure, normal VAD sounds, peripheral edema and mean arterial pressure of 88mmHg. Monitoring of the LVAD demonstrated abnormal waveforms suggestive of intermittent suction events (Panel 1A). The patient was transferred to the ICU for invasive monitoring and transesophageal echocardiography to allow a ramp study. The imaging and hemodynamic data confirmed occlusion of the LVAD inflow cannula on each cardiac cycle (B,C) at his existing VAD settings (2740 rpm). A ramp study revealed a paradoxical increase in right atrial pressure and no improvement in blood pressure at higher speeds (D). Thoracic CT imaging confirmed cannula malposition and review of LVAD log files determined that there had been an abrupt change in pulsatility two months after implant and there had been increasingly frequent suction events since. Surgical reintervention was considered, however, given the chronicity of the changes, the consensus was to list the patient for urgent transplantation. <h3>Summary</h3> We describe a case of covert suction events where increasing body mass by 25kg (41%) may have contributed. In LVAD patients who are challenging to decongest, waveforms, hemodynamic and imaging guided optimization can improve function and identify complications.

Optimizing early cancer diagnosis and detection using a temporal subtraction technique

To optimize the early diagnosis and detection of lung cancer, computer-aided diagnostic (CAD) systems have been a useful tool for analyzing medical images. The temporal subtraction technique, which is a CAD system, performs the subtraction operation between the current image and the previous image on the same patient, and supports observation by emphasizing the temporal changes. However, the temporal subtraction technique for 3D images, such as thoracic CT images, has not yet been established. There is a need to develop efficient and highly accurate 3D nonrigid registration techniques to reduce subtraction artifacts. This study aims to develop a 3D nonrigid registration technique to establish a 3D temporal subtraction technique. In particular, we focus on the Finite Element Method, which is versatile, applicable to a wide range of fields, and capable of handling any shape. Our new method was examined on 46 clinical cases with multidetector row computed tomography images. As a result, the proposed method improved by 6.93% (p = 3.0 × 10−6) compared to the conventional methods in terms of the rate of reduction of artifacts, and the effectiveness was verified. Therefore, this study contributes to the literature on early detection and treatment.

SVM による胸部CT 画像からのドライバー遺伝子情報変異の検出

SVM による胸部CT 画像からのドライバー遺伝子情報変異の検出

Forefront of AI Applications for COVID-19 Imaging Diagnosis

The intra- and inter-observer variability in diagnosis of thoracic CT images may affect the diagnosis of COVID-19. Therefore, several studies have been reported to develop artificial intelligence (AI) approaches using deep learning (DL) and radiomics technologies. The difference between them is automatic feature extraction (DL) and hand-crafted one (radiomics). The advantages of the AI-based imaging approaches for the COVID-19 are fast throughput, non-invasion, quantification, and integration of PCR results, CT findings, and clinical information. To the best of my knowledge, three types of the AI approaches have been studied: detection, severity differentiation, and prognosis prediction of COVID-19. AI technologies on assessment of severity/prediction of prognosis for COVID-19 may be more crucial than detection of COVID-19 pneumonia after COVID-19 becomes one of common diseases.

Use of routine computed tomography scans for detecting osteoporosis in thoracolumbar vertebral bodies.

This study observed the distribution of CT attenuation values for T10-L3 vertebral bodies and derived the Hounsfield unit (HU) thresholds using the quantitative computed tomography (QCT) as a reference to predict osteoporosis and normal bone density. We included 482 subjects who were scheduled to undergo CT lung cancer screening and pulmonary nodule follow-up from May 2015 to February 2019. The subjects were scanned with the calibration phantom beneath the back while performing a chest CT scan. The volumetric bone mineral density (vBMD) and CT attenuation values of T10-L3 vertebral bodies were measured, and the correlation between the two measurements was analyzed. Receiver operator characteristic (ROC) curves were generated to determine diagnostic optimal thresholds. A total of 2716 vertebral bodies of 457 participants were measured after exclusion screening. CT attenuation value of each plane's vertebral body showed a strong correlation with vBMD. The optimal threshold of > 141HU was 93.5% sensitive and 86.1% specific for the recognition of normal BMD. The optimal threshold of < 102.4HU was 96.9% specific and 82.1% sensitive for distinguishing osteoporosis from osteopenia and normal BMD. The average CT attenuation values of vertebral bodies with compressed and normal morphology were 108.9 ± 20.6 and 136.8 ± 32.2HU, respectively. Sagittal reconstruction of the thoracic vertebrae using routine thoracic CT image combined with CT attenuation value measurements of the spine is valuable for predicting bone mineral density in high-risk populations. The mean CT attenuation values of the vertebral bodies with vertebral compression appearance were lower than that of normal vertebral shape.

Evaluation of Azygos Vein to Aorta Ratio using Multi-Detector Computed Tomography in Dogs

The purpose of this study is to evaluate the azygos vein to aorta diameter ratio (AV/AO ratio) in healthy dogs and dogs with diseases that may cause azygos vein dilation. The normal groups included eleven healthy beagle dogs and eleven small breed dogs. The abnormal group included 43 dogs that had diseases with high right atrial pressure (RA group), obstruction of cranial or caudal vena cava (CVC group) and porto-azygos shunt group (PSS group). The diameter of AV and AO were measured at three sites on thoracic CT images; the level of the cranial endplate of 12th thoracic vertebra (TV level), the tracheal bifurcation (TB level) and the insertion to the cranial vena cava (CrVC level). The AV/AO ratio at the CrVC level had significantly higher values than those at the TV and TB levels in the normal groups. The AV/AO ratios of TV and TB levels in the CVC group and the values at all measurement sites of the PSS group were significantly higher than the normal groups. AV/AO ratio will be a useful factor for the evaluation of various clinical conditions that change the AV diameter with respect independent to the body weight in dogs.

Pulmonary trunk to ascending aorta ratio and reference values for diameters of pulmonary arteries and main bronchi in healthy adults

Objectives: The ratio of the diameter of pulmonary trunk (PT) to the diameter of the ascending aorta (AA) is used to evaluate cardiopulmonary diseases. Different values have been reported for the normal value of PT:AA ratio (to be less than 0.9, 1 or 1.4). In this study, we aimed to investigate the diameters of the PT, right (RPA) and left pulmonary artery (LPA), AA, right (RMB) and left main bronchus LMB) using multidetector computed tomography (MDCT) and to determine reference values for PT:AA according to age and sex in normal healthy adults. Methods: Thoracic CT images of 200 individuals, (103 males, 97 females; age 18–89 years), without cardiopulmonary pathology and surgery, were retrospectively evaluated using MDCT. Diameters of PT, RPA, LPA, AA were measured at the level of the pulmonary artery bifurcationand PT:AA ratio was calculated. Results: The mean diameters of PT, AA, RPA, LPA, RMB, LMB were found as 2.7±0.51 cm, 3.25±0.63 cm, 1.98±0.46 cm, 1.81±0.43 cm, 1.73±0.35 cm, and 1.66±0.55 cm, respectively. The mean value of PT:AA ratio was 0.84±0.18 cm in males and 0.86±0.13 cm in females. Conclusion: Determining the normal values of related measurements will contribute to diagnosis and treatment of cardiopulmonary diseases.

PENGGUNAAN VARIASI FILTER PADA WINDOWING LUNG PADA PEMERIKSAAN CT SCAN THORAX

Background: Filter is an image processing parameter applied to raw data used to determine the attenuation value of each pixel of CT Scan imaging, and it serves to minimize the occurrence of star patterned streak artifacts formed on the CT images. The use of typical Y-sharp or Smooth filters is common for the thoracic CT imaging in the clinic. Yet, the studies to compare which of the typical filters that provide a good imaging technique were clinically limited to find in practices. The purpose of this study is to know wheter there is a difference or not in the anatomical image resolved amongst the thoracic CT images when using the Y-Sharp and Smooth filters; and to determine which of the filter that produces the better image quality by comparison.Methods: This is quantitative research with approached the pre-experimental design with the static-groups of comparison. Image data were taken from the raw data of the 10 patients who underwent CT Scan Thorax examinations. All the image data were reconstructed by means of using the Y-Sharp and Smooth filters. Assessment of the images were performed by 2 radiologists. The Cohen's Kappa was used for image evaluators agreement where the Wilcoxon Test was applied to test the hypotheses.Result: The result of this study showed that there is a significant difference in anatomical organ visualization of the thoracic CT images between filter the Y-Sharp and Smooth filters with p-value 0.005. The use of Y-Sharp filter is better as it produces a high spatial image resolution.Conclusion: The anatomical organs of the thoracic CT images are visually reproduced when the Y-sharp filter is employed. The comparison between the two different filter uses deems significance.

Application of radiomics signature captured from pretreatment thoracic CT to predict brain metastases in stage III/IV ALK-positive non-small cell lung cancer patients.

The purpose of this study is to develop a radiomics approach to predict brain metastasis (BM) for stage III/IV ALK-positive non-small cell lung cancer (NSCLC) patients. Patients with ALK-positive III/IV NSCLC from 2014 to 2017 were enrolled retrospectively. Their pretreatment thoracic CT images were collected, and the gross tumor volume (GTV) was defined by two experienced radiation oncologists. An in-house feature extraction code-set was performed based on MATLAB 2015b (Mathworks, Natick, MA, USA) in patients' CT images to extract features. Patients were randomly divided into training set and test set (4:1) by using createDataPartition function in caret package. A test-retest in RIDER NSCLC dataset was performed to identify stable radiomics features. LASSO Cox regression and a leave-one-out cross-validation were conducted to identify optimal features for the logistic regression model to evaluate the predictive value of radiomics feature(s) for BM. Furthermore, extended validation for the radiomics feature(s) and Cox regression analyses which combined radiomics feature(s) and treatment elements were implemented to predict the risk of BM during follow-up. In total, 132 patients were included, among which 27 patients had pretreatment BM. The median follow-up time was 11.8 (range, 0.1-65.2) months. In the training set, one radiomics feature (W_GLCM_LH_Correlation) showed discrimination ability of BM (P value =0.014, AUC =0.687, 95% CI: 0.551-0.824, specificity =83.5%, sensitivity =57.1%). It also exhibited reposeful performance in the test set (AUC =0.642, 95% CI: 0.501-0.783, specificity =60.0%, sensitivity =83.3%). Those 105 patients without pretreatment BM were divided into stage III (n=57) and stage IV (n=48) groups. The radiomics feature (W_GLCM_LH_Correlation) had moderate performance to predict BM during/after treatment in separate groups (stage III: AUC =0.682, 95% CI: 0.537-0.826, specificity =64.4%, sensitivity =75.0%; stage IV: AUC =0.653, 95% CI: 0.503-0.804, specificity =70.4%, sensitivity =75.0%). Meanwhile, stage III patients could be divided into low risk and high risk groups for BM during surveillance according to Cox regression analysis (log-rank P value =0.021). We identified one wavelet texture feature derived from pretreatment thoracic CT that presented potential in predicting BM in stage III/IV ALK-positive NSCLC patients. This could be beneficial to risk stratification for such patients. Further investigation is necessary to include expanded sample size investigation and external multicenter validation.

Torso injuries after fall from standing-empiric abdominal or thoracic CT imaging is not indicated

PurposeFalls from standing (FFS) have become the most common mechanism of injury at many trauma centers. Liberal imaging of low energy trauma has questionable value. We hypothesize that torso trauma intervention is rare in the FFS population, and physical examination sufficiently screens for torso injuries needing intervention. MethodsWe queried our ACS-verified Level 1 trauma center registry for falls from standing from 1/1/14 to 12/31/16. Exclusion criteria were: falls from height, falls associated with penetrating trauma, lack of an abdominal or chest CT, a Glasgow Coma Scale Score (GCS) less than 15, and surgical intervention at another facility prior to arrival at our center. Demographics, historical details, hemodynamics, injuries, injury severity, procedures, initial vital signs, and outcome were recorded. Results1,654 patients had a FFS during our study period. 728 had an abdominal or chest CT and a GCS of 15 and comprised the evaluable population. Mean age was 56.5 years. 55.8% were female. The mortality rate was 8%. There were 179 chest injuries in 121 patients, and 54 abdominal injuries in 43 patients. 379 patients had a GCS of 15 and underwent thoracic CT, yet only 11 (3%) underwent intervention. The negative predictive value for physical exam was 100% for chest intervention. 349 patients had a GCS of 15 and abdominal CT, yet only 13 (3.7%) underwent procedural intervention. Abdominal physical exam had a negative predictive value of 99.7% for intervention, but when combined with vital signs, the value was 100%. ConclusionTorso injuries in FFS are rare. Of our study population, 13 abdominal injuries underwent intervention, and 11 chest injuries underwent intervention. Screening patients by physical examination and vital signs is sufficient and safely allows for the use of selective abdominal and chest CT.

Assessment of patient dose and noise level of clinical CT images: automated measurements

We investigated comparisons between patient dose and noise in pelvic, abdominal, thoracic and head CT images using an automatic method. 113 patient images (37 pelvis, 34 abdominal, 25 thoracic, and 17 head examinations) were retrospectively and automatically examined in this study. Water-equivalent diameter (Dw), size-specific dose estimates (SSDE) and noise were automatically calculated from the center slice for every patient image. The Dw was calculated based on auto-contouring of the patients’ edges, and the SSDE was calculated as the product of the volume CT dose index (CTDIvol) extracted from the Digital Imaging and Communications in Medicine (DICOM) header and the size conversion factor based on the Dw obtained from AAPM 204. The noise was automatically measured as a minimum standard deviation in the map of standard deviations. A square region of interest of about 1 cm2 was used in the automated noise measurement. The SSDE values for the pelvis, abdomen, thorax, and head were 21.8 ± 7.3 mGy, 22.0 ± 4.5 mGy, 21.5 ± 4.7 mGy, and 65.1 ± 1.7 mGy, respectively. The SSDEs for the pelvis, abdomen, and thorax increased linearly with increasing Dw, and for the head with constant tube current, the SSDE decreased with increasing Dw. The noise in the pelvis, abdomen, thorax, and head were 5.9 ± 1.5 HU, 5.2 ± 1.4 HU, 4.9 ± 0.8 HU and 3.9 ± 0.2 HU, respectively. The noise levels for the pelvis, abdomen, and thorax of the patients were relatively constant with Dw because of tube current modulation. The noise in the head image was also relatively constant because Dw variations in the head are very small. The automated approach provides a convenient and objective tool for dose optimizations.

Computer-Aided Diagnosis (CAD) of Pulmonary Nodule of Thoracic CT Image Using Transfer Learning.

Computer-aided diagnosis (CAD) has already been widely used in medical image processing. We recently make another trial to implement convolutional neural network (CNN) on the classification of pulmonary nodules of thoracic CT images. The biggest challenge in medical image classification with the help of CNN is the difficulty of acquiring enough samples, and overfitting is a common problem when there are not enough images for training. Transfer learning has been verified as reasonable in dealing with such problems with an acceptable loss value. We use the classic LeNet-5 model to classify pulmonary nodules of thoracic CT images, including benign and malignant pulmonary nodules, and different malignancies of the malignant nodules. The CT images are obtained from Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) where both pulmonary nodule scanning and nodule annotations are available. These images are labeled and stored in a medical images knowledge base (KB), which is designed and implemented in our previous work. We implement the 10-folder cross validation (CV) to testify the robustness of the classification model we trained. The result demonstrates that the transfer learning of the LeNet-5 is good for classifying pulmonary nodules of thoracic CT images, and the average values of Top-1 accuracy are 97.041% and 96.685% respectively. We believe that our work is beneficial and has potential for practical diagnosis of lung nodules.

Predictors of Response to Endobronchial Coil Therapy in Patients With Advanced Emphysema

The Lung Volume Reduction Coil Treatment in Patients With Emphysema (RENEW) trial reported improvements in quality of life, pulmonary function, and exercise performance following endobronchial coil treatment. The purpose of this post hoc analysis was to identify baseline predictors, including quantitative CT measures, that identify patients most likely to significantly benefit from endobronchial coil therapy. Quantitative CT analysis by an independent radiology laboratory and a qualitative evaluation by five blinded experts of the baseline thoracic CT imaging were performed. Univariate and multivariate logistic regression analyses were performed to elucidate characteristics associated with clinical response. In total, 125 patients underwent coil treatment and had evaluable 12-month follow-up results. Of these, 78 patients received treatment of lobes with the highest emphysematous destruction determined by quantitative CT analysis (quantitative visual match [QVM]+), and 47 received treatment in at least one lobe that was not the most destroyed (QVM-). From the 78 patients with QVM+ treatment, a subgroup of 50 patients (64%) was identified with baseline residual volume > 200%predicted, emphysema score > 20%low attenuation area, and absence of airway disease. In this subgroup, greater lobar residual volume reduction in the treated lobes was achieved, which was associated with significant mean ± SE improvement in FEV1 (15.2 ± 3.1%), St. George's Respiratory Questionnaire (-12 ± 2points), and residual volume (-0.57 ± 0.13 L). This post hoc analysis found that both significant hyperinflation (residual volume≥ 200%predicted) and CT analysis are critical for patient selection and treatment planning for endobronchial coil therapy. Quantitative CT analysis is important to identify optimal lobar treatment and to exclude patients with insufficient emphysema (< 20%low attenuation area), whereas visual assessment identifies patients with signs of airway disease associated with worse outcomes. ClinicalTrials.gov; No.: NCT01608490; URL: www.clinicaltrials.gov.

CT-Defined Body Composition in Adult Patients with Hematological Malignancies Prior to Allogeneic Hematopoietic Stem Cell Transplantation (alloHSCT): Endemic Sarcopenia and Reduced Muscle Radiodensity Associate with Impaired Forced Expiratory Volume (FEV)1

Background Precise quantification of skeletal muscle using computed tomography is accessible using cancer patients' standard oncologic images. Reduced muscle mass (i.e. sarcopenia) has been related to mortality, treatment complications and QOL and reduced respiratory muscle strength, however these associations are not well characterized in adult alloHSCT recipients. Methods A consecutive retrospective series (n=296) patients who had allogeneic BMT at a comprehensive Center between 01/2016 and 12/2017. Pre-transplant CT scans were used to quantify skeletal muscle and adipose tissue at the 3rd lumbar (L3) and/or 4th thoracic (T4) vertebra. Tumor and patient characteristics were recorded, including FEV1 by spirometry. Sarcopenia was defined according to Prado et al. (PMID:18539529) Results 296 patients (male n=161; female n=135) were included, all of whom had thoracic-CT; a subset of these (n=195) also had lumbar-CT. Diagnoses were NHL (n=172), AML (n=66) HD (n=14), ALL (n=14), MDS (n=18) and other (MM, MPN, CML; n=12). Patients had increased BMI (male= 28.7 kg/m2, female=26.5 kg/m2), with many patients (♂64.7%; ♀57.0%) meeting criteria for sarcopenia. Rates of sarcopenia were higher (p Men and women were more muscular at T4 than L3. L3 and T4 skeletal muscle areas were moderately correlated (r2 =0.33- 0.47), as were L3 and T4 muscle radiodensities (r2 =0.58-0.63) and subcutaneous fat areas (r2 =0.69- 0.71), (p In multivariable linear regression, independent predictors of muscularity (i.e. T4 muscle index) were age (p=0.005), sex (p=0.000), BMI (p=0.000) and HCT-CI (p=0.057). Disease severity (Armand DRI) (p=0.131) was not significantly associated with T4 muscle index. Similar conclusions were obtained for L3 muscle index. In multivariable linear regression adjusted for sex (p=0.000), age (p=0.000) and HCT-CI (p=0.000) and both T4 muscle index (b 0.252 [95%CI 0.126;0.412] p=0.000) and T4 muscle radiodensity (b 0.165 [95%CI 0.110;0.630] p=0.006) were independently associated with FEV1; Armand DRI (p=0.712) and KPS (p=0.991) were not associated with FEV1. Similar conclusions were obtained when muscle index and radiodensity at L3 were considered. Discussion and Conclusion Lumbar or thoracic CT images are useful for body composition assessment in this population, and reveal high rates of sarcopenia, similar to those reported in very elderly patients. Reduced muscle mass and radiodensity associate with impaired FEV1. More muscular patients were male, younger, of high BMI and had fewer comorbidities.

Toward an Affine Feature-Based Registration Method for Ground Glass Lung Nodule Tracking.

Lung nodule progression assessment from medical imaging is a critical biomarker for assessing the course of the disease or the patient's response to therapy. CT images are routinely used to identify the location and size and rack the progression of lung nodules. However, nodule segmentation is challenging and prone to error, due to the irregular nodule boundaries, therefore introducing error in the lung nodule quantification process. Here, we describe the development and evaluation of a feature-based affine image registration framework that enables us to register two time point thoracic CT images as a means to account for the back-ground lung tissue deformation, then use digital subtraction images to assess tumor progression/regression. We have demonstrated this method on twelve de-identified patient datasets and showed that the proposed method yielded a better than 1.5mm registration accuracy vis-à-vis the widely accepted non-rigid image registration techniques. To demonstrate the potential clinical value of our described technique, we conducted a study in which our collaborating clinician was asked to provide an assessment of nodule progression/regression using the digital subtraction images post-registration. This assessment was consistent, yet provided more confidence, than the traditional lung nodule tracking based on visual analysis of the CT images.

A SIMPLIFIED METHOD FOR THE WATER-EQUIVALENT DIAMETER CALCULATION TO ESTIMATE PATIENT DOSE IN CT EXAMINATIONS.

We proposed and evaluated a water-equivalent diameter calculation without using a region of interest (ROI), (Dw,t) and compared it with the results of using a ROI fitted to the patient border (Dw,f). Evaluations were carried out on thoracic and head CT images. We found that the difference between Dw,t and Dw,f was within 5% for all images in the head region, and most images were within 5% (27 of the 30 patients, 90%) in the thoracic region. We also proposed a method to automatically detect and eliminate the patient table (or head support) from images and evaluated the water-equivalent diameter values after the table had been removed (Dw,nt). This method was able to recognize and remove the patient table from all images used. By removing the table, the water-equivalent diameter (Dw,nt) became more accurate and the difference from Dw,f was within 5% for all images (head and thoracic images).

Ultra‐low‐dose CT image denoising using modified BM3D scheme tailored to data statistics

It is important to enhance image quality for low-dose CT acquisitions to push the ALARA boundary. Current state-of-the-art block-matching three-dimensional (BM3D) denoising scheme assumes white Gaussian noise (WGN) model. This study proposes a novel filtering module to be incorporated into the BM3D framework for ultra-low-dose CT denoising, by accounting for its specific power spectral properties. In the current BM3D algorithm, the Wiener filtering is applied in the transform domain to a post-thresholding signal for enhanced denoising. However, unlike most natural/synthetic images, low-dose CTs do not obey the ideal Gaussian noise model. Based on the specific noise properties of ultra-low-dose CT, we derive the optimal transform-domain coefficients of Wiener filter based on the minimum mean-square-error (MMSE) criterion, taking the noise spectrum and the signal/noise cross spectrum into consideration. In the absence of ground-truth signal, the hard-thresholding denoising module in the previous stage is used as a plug-in estimator. We evaluate the denoising performance on thoracic CT image datasets containing paired full-dose and ultra-low-dose images simulated by a well-validated clinical engine (or pipeline). We also assess its clinical implication by applying the denoising methods to the emphysema quantification task. Our modified BM3D method is compared with the current one, using peak signal-to-noise ratio (PSNR) and emphysema scoring results as evaluation metrics. The noise in ultra-low-dose CT presented distinct non-Gaussian characteristics and was correlated with image intensity. Performance evaluation showed that the current Wiener filter in basic BM3D algorithm yielded little denoising enhancement on ultra-low-dose CT images. In contrast, the proposed Wiener filter achieved (1.46, 1.91)dB performance gain in mean and median peak signal-to-noise ratio (PSNR) for 5%-dose image denoising and (0.93, 0.95)dB improvement for 10% dose. A paired t-test of the PSNRs between denoising using the current and the proposed Wiener filters demonstrated statistically significant improvement, yielding P-values of 1.45E-12 and 1.34E-7 on 5% and 10%-dose images, respectively. In addition, emphysema quantification on the denoised images using the modified BM3D method also had statistically significant advantage over that using the current BM3D scheme, resulting in a P-value of 6.30E-5 with the commonly used measure. This work tailors the Wiener filter in BM3D algorithm to data statistics and demonstrates statistically significant performance improvement on ultra-low-dose CT image denoising and a subsequent emphysema quantification task. Such performance gain is more pronounced with a lower dose level. The development and rationale are generally enough for other image denoising tasks when the WGN assumption is violated.

Behaviour and motor performance in a cognitive-motor task during learning processes: evidence from EEG network analysis

Thoracic endovascular aortic repair (TEVAR) has dramatically expanded treatment options for patients with thoracic aortic pathology. The interaction between endografts and the dynamic anatomy of the thoracic aorta is not well characterized for repetitive physiologic stressors and subsequent issues related to long-term durability. Through three-dimensional (3D) modeling we sought to quantify cardiac-induced aortic deformation before and after TEVAR to assess the impact of endografts on dynamic aortic anatomy.Eight patients with acute (n = 4) or chronic (n = 3) type B dissections, or chronic arch aneurysm (n = 1), underwent TEVAR with a single (n = 5) or multiple (n = 3) Gore C-TAG(s). Cardiac-resolved thoracic CT images were acquired pre- and post-TEVAR. 3D models of thoracic aorta and branch vessels were constructed in systole and diastole. Axial length, mean, and peak curvature of the ascending aorta, arch, and stented lumens were computed from the aortic lumen centerline, delineated with branch vessel landmarks. Cardiac-induced deformation was computed from mid-diastole to end-systole.Pre-TEVAR, there were no significant cardiac-induced changes for aortic axial length or mean curvature. Post-TEVAR, the ascending aorta increased in axial length (2.7 ± 3.1%, P < 0.05) and decreased in mean curvature (0.38 ± 0.05 → 0.36 ± 0.05 cm−1, P < 0.05) from diastole to systole. From pre- to post-TEVAR, axial length change increased in the ascending aorta (P < 0.02), mean curvature decreased in the arch and stented aorta (P < 0.03), and peak curvature decreased in the stented aorta (P < 0.05).TEVAR for a range of indications not only causes direct geometric changes to the stented aorta but also results in dynamic changes to the ascending and stented aorta. In our cohort, endograft placement straightens the stented aorta and mutes cardiac-induced bending due to longitudinal stiffness. This is compensated by greater length and curvature changes from diastole to systole in the ascending aorta, relative to pre-TEVAR.

Iterative inversion of deformation vector fields with feedback control.

Often, the inverse deformation vector field (DVF) is needed together with the corresponding forward DVF in four-dimesional (4D) reconstruction and dose calculation, adaptive radiation therapy, and simultaneous deformable registration. This study aims at improving both accuracy and efficiency of iterative algorithms for DVF inversion, and advancing our understanding of divergence and latency conditions. We introduce a framework of fixed-point iteration algorithms with active feedback control for DVF inversion. Based on rigorous convergence analysis, we design control mechanisms for modulating the inverse consistency (IC) residual of the current iterate, to be used as feedback into the next iterate. The control is designed adaptively to the input DVF with the objective to enlarge the convergence area and expedite convergence. Three particular settings of feedback control are introduced: constant value over the domain throughout the iteration; alternating values between iteration steps; and spatially variant values. We also introduce three spectral measures of the displacement Jacobian for characterizing a DVF. These measures reveal the critical role of what we term the nontranslational displacement component (NTDC) of the DVF. We carry out inversion experiments with an analytical DVF pair, and with DVFs associated with thoracic CT images of six patients at end of expiration and end of inspiration. The NTDC-adaptive iterations are shown to attain a larger convergence region at a faster pace compared to previous nonadaptive DVF inversion iteration algorithms. By our numerical experiments, alternating control yields smaller IC residuals and inversion errors than constant control. Spatially variant control renders smaller residuals and errors by at least an order of magnitude, compared to other schemes, in no more than 10 steps. Inversion results also show remarkable quantitative agreement with analysis-based predictions. Our analysis captures properties of DVF data associated with clinical CT images, and provides new understanding of iterative DVF inversion algorithms with a simple residual feedback control. Adaptive control is necessary and highly effective in the presence of nonsmall NTDCs. The adaptive iterations or the spectral measures, or both, may potentially be incorporated into deformable image registration methods.