Association between the Static and Dynamic Lung Function and CT-Derived Thoracic Skeletal Muscle Measurements-A Retrospective Analysis of a 12-Month Observational Follow-Up Pilot Study.

BACKGROUND: Patients with chronic obstructive pulmonary disease (COPD) with low skeletal muscle mass and severe airway obstruction have higher mortality risks. However, the relationship between dynamic/static lung function (LF) and thoracic skeletal muscle measurements (SMM) remains unclear. This study explored patient characteristics (weight, BMI, exacerbations, dynamic/static LF, sex differences in LF and SMM, and the link between LF and SMM changes. METHODS: A retrospective analysis of a 12-month prospective follow-up study patients with stable COPD undergoing standardized treatment, covering mild to severe stages, was conducted. The baseline and follow-up assessments included computed tomography and body plethysmography. RESULTS: This study included 35 patients (17 females and 18 males). This study revealed that females had more stable LF but tended to have greater declines in SMM areas and indices than males (-5.4% vs. -1.9%, respectively), despite the fact that females were younger and had higher LF and less exacerbation than males. A multivariate linear regression showed a negative association between the inspiratory capacity/total lung capacity ratio (IC/TLC) and muscle fat area. CONCLUSIONS: The findings suggest distinct LF and BC progression patterns between male and female patients with COPD. A low IC/TLC ratio may predict increased muscle fat. Further studies are necessary to understand these relationships better.

Ethnic differences in CT derived abdominal body composition measures: a comparative retrospect pilot study between European and Inuit study population.

Understanding ethnic variations in body composition is crucial for assessing health risks. Universal models may not suit all ethnicities, and there is limited data on the Inuit population. This study aimed to compare body composition between Inuit and European adults using computed tomography (CT) scans and to investigate the influence of demographics on these measurements. A retrospective analysis was conducted on 50 adults (29 Inuit and 21 European) who underwent standard trauma CT scans. Measurements focused on skeletal muscle index (SMI), various fat indices, and densities at the third lumbar vertebra level, analyzed using the Wilcoxon-Mann-Whitney test and multiple linear regression. Inuit women showed larger fat tissue indices and lower muscle and fat densities than European women. Differences in men were less pronouncehd, with only Intramuscular fat density being lower among Inuit men. Regression indicated that SMI was higher among men, and skeletal muscle density decreased with Inuit ethnicity and age, while visceral fat index was positively associated with age. This study suggests ethnic differences in body composition measures particularly among women, and indicates the need for Inuit-specific body composition models. It higlights the importance of further research into Inuit-specific body composition measurements for better health risk assessment.

Association between thoracic and third lumbar CT-derived muscle mass and density in Caucasian patients without chronic disease: a proof-of-concept study.

BACKGROUND: Computed tomography (CT) is increasingly used in the clinical workup, and existing scan contains unused body composition data, potentially useful in a clinical setting. However, there is no healthy reference for contrast-enhanced thoracic CT-derived muscle measures. Therefore, we aimed at investigating whether there is a correlation between each of the thoracic and third lumbar vertebra level (L3) skeletal muscle area (SMA), skeletal muscle index (SMI), and skeletal muscle density (SMD) at contrast-enhanced CT in patients without chronic disease. METHODS: A proof-of-concept retrospective observational study was based on Caucasian patients without chronic disease, who received CT for trauma between 2012 and 2014. Muscle measures were assessed using a semiautomated threshold-based software by two raters independently. Pearson's correlation between each thoracic level and third lumbar and intraclass correlation between two raters and test-retest with SMA as proxy parameters were used. RESULTS: Twenty-one patients (11 males, 10 females; median age 29 years) were included. The second thoracic vertebra (T2) had the highest median of cumulated SMA (males 314.7 cm2, females 118.5 cm2) and SMI (97.8 cm2/m2 and 70.4 cm2/m2, respectively). The strongest SMA correlation was observed between T5 and L3 (r = 0.970), the SMI between T11 and L3 (r = 0.938), and the SMD between the T10 and L3 (r = 0.890). CONCLUSIONS: This study suggests that any of the thoracic levels can be valid to assess skeletal muscle mass. However, the T5 may be most favourable for measuring SMA, the T11 for SMI, and T10 for SMD when using contrast-enhanced thoracic CT. RELEVANCE STATEMENT: In COPD patients, a CT-derived thoracic muscle mass assessment may help identify who would benefit from focused pulmonary rehabilitation: thoracic contrast-enhanced CT conducted as part of the standard clinical workup can be used for this evaluation. KEY POINTS: • Any thoracic level can be used to assess thoracic muscle mass. • Thoracic level 5 is strongly associated with the 3rd lumbar muscle area. • A strong correlation between the thoracic level 11 and the 3rd lumbar muscle index. • Thoracic level 10 is strongly associated with the 3rd lumbar muscle density.

Computer-Assisted Annotation of Digital H&E/SOX10 Dual Stains Generates High-Performing Convolutional Neural Network for Calculating Tumor Burden in H&E-Stained Cutaneous Melanoma.

Deep learning for the analysis of H&E stains requires a large annotated training set. This may form a labor-intensive task involving highly skilled pathologists. We aimed to optimize and evaluate computer-assisted annotation based on digital dual stains of the same tissue section. H&E stains of primary and metastatic melanoma (N = 77) were digitized, re-stained with SOX10, and re-scanned. Because images were aligned, annotations of SOX10 image analysis were directly transferred to H&E stains of the training set. Based on 1,221,367 annotated nuclei, a convolutional neural network for calculating tumor burden (CNNTB) was developed. For primary melanomas, precision of annotation was 100% (95%CI, 99% to 100%) for tumor cells and 99% (95%CI, 98% to 100%) for normal cells. Due to low or missing tumor-cell SOX10 positivity, precision for normal cells was markedly reduced in lymph-node and organ metastases compared with primary melanomas (p < 0.001). Compared with stereological counts within skin lesions, mean difference in tumor burden was 6% (95%CI, -1% to 13%, p = 0.10) for CNNTB and 16% (95%CI, 4% to 28%, p = 0.02) for pathologists. Conclusively, the technique produced a large annotated H&E training set with high quality within a reasonable timeframe for primary melanomas and subcutaneous metastases. For these lesion types, the training set generated a high-performing CNNTB, which was superior to the routine assessments of pathologists.

Sex differences in microvascular function across lower leg muscles in humans.

Studies have reported sex-based differences in conduit artery function, however little is known about possible sex-based differences in microvascular function, and possible influence of muscle group. Blood-oxygen-level-dependent (BOLD) MR images acquired during ischemia-reperfusion assess the reactive hyperemic response in the microvasculature of skeletal muscle. We tested the hypothesis that women have greater microvascular reactivity, reflected by faster time-to-peak (TTP) and time-to-half-peak (TTHP) of the BOLD response, across all lower leg muscles. In healthy, young men (n = 18) and women (n = 12), BOLD images of both lower legs were acquired continuously during 30 s of rest, 5 min of cuff occlusion and 2 min of reperfusion, in a 3 T MR scanner. Segmentation of tibialis anterior (TA), soleus (SO), gastrocnemius medial (GM), and the peroneal group (PG) were performed using image registration, and TTP and TTHP of the BOLD response were determined for each muscle. Overall, women had faster TTP (p = 0.001) and TTHP (p = 0.01) than men. Specifically, women had shorter TTP and TTHP in TA (27.5-28.4%), PG (33.9-41.6%), SO (14.7-19.7%) and GM (15.4-18.8%). Overall, TTP and TTHP were shorter in TA compared with PG (25.1-31.1%; p ≤ 0.007), SO (14.3-16%; p ≤ 0.03) and GM (15.6-26%; p ≤ 0.01). Intra class correlations analyses showed large variation in absolute agreement (range: 0.10-0.81) of BOLD parameters between legs (within distinct muscles). Faster TTP and TTHP across all lower leg muscles, in women, provide novel evidence of sex-based differences in microvascular function of young adults matched for age, body mass index, and physical activity level.

The global end-ranges of neck flexion and extension do not represent the maximum rotational ranges of the cervical intervertebral joints in healthy adults - an observational study.

BACKGROUND: In clinical diagnosis, the maximum motion of a cervical joint is thought to be found at the joint's end-range and it is this perception that forms the basis for the interpretation of flexion/extension imaging studies. There have however, been representative cases of joints producing their maximum motion before end-range, but this phenomenon is yet to be quantified. PURPOSE: To provide a quantitative assessment of the difference between maximum joint motion and joint end-range in healthy subjects. Secondarily to classify joints into type based on their motion and to assess the proportions of these joint types. STUDY DESIGN: This is an observational study. SUBJECT SAMPLE: Thirty-three healthy subjects participated in the study. OUTCOME MEASURES: Maximum motion, end-range motion and surplus motion (the difference between maximum motion and end-range) in degrees were extracted from each cervical joint. METHODS: Thirty-three subjects performed one flexion and one extension motion excursion under video fluoroscopy. The motion excursions were divided into 10% epochs, from which maximum motion, end-range and surplus motion were extracted. Surplus motion was then assessed in quartiles and joints were classified into type according to end-range. RESULTS: For flexion 48.9% and for extension 47.2% of joints produced maximum motion before joint end-range (type S). For flexion 45.9% and for extension 46.8% of joints produced maximum motion at joint end-range (type C). For flexion 5.2% of joints and for extension 6.1% of joints concluded their motion anti-directionally (type A). Significant differences were found for C2/C3 (P = 0.000), C3/C4 (P = 0.001) and C4/C5 (P = 0.005) in flexion and C1/C2 (P = 0.004), C3/C4 (P = 0.013) and C6/C7 (P = 0.013) in extension when comparing the joint end- range of type C and type S. The average pro-directional (motion in the direction of neck motion) surplus motion was 2.41° ± 2.12° with a range of (0.07° -14.23°) for flexion and 2.02° ± 1.70° with a range of (0.04°-6.97°) for extension. CONCLUSION: This is the first study to categorise joints by type of motion. It cannot be assumed that end-range is a demonstration of a joint's maximum motion, as type S constituted approximately half of the joints analysed in this study.

Pancreatic magnetic resonance imaging texture analysis in chronic pancreatitis: a feasibility and validation study.

PURPOSE: This feasibility and validation study addresses the potential use of magnetic resonance imaging (MRI) texture analysis of the pancreas in patients with chronic pancreatitis (CP). METHODS: Extraction of 851 MRI texture features from diffusion weighted imaging (DWI) of the pancreas was performed in 77 CP patients and 22 healthy controls. Features were reduced to classify patients into subgroups, and a Bayes classifier was trained using a tenfold cross-validation forward selection procedure. The classifier was optimized to obtain the best average m-fold accuracy, sensitivity, specificity, and positive predictive value. Classifiers were: presence of disease (CP vs. healthy controls), etiological risk factors (alcoholic vs. nonalcoholic etiology of CP and tobacco use vs. no tobacco use), and complications to CP (presumed pancreatogenic diabetes vs. no diabetes and pancreatic exocrine insufficiency vs. normal pancreatic function). RESULTS: The best classification performance was obtained for the disease classifier selecting only five of the original features with 98% accuracy, 97% sensitivity, 100% specificity, and 100% positive predictive value. The risk factor classifiers obtained good performance using 9 (alcohol: 88% accuracy) and 10 features (tobacco: 86% accuracy). The two complication classifiers obtained similar accuracies with only 4 (diabetes: 83% accuracy) and 3 features (exocrine pancreatic function: 82% accuracy). CONCLUSION: Pancreatic texture analysis demonstrated to be feasible in patients with CP and discriminate clinically relevant subgroups based on etiological risk factors and complications. In future studies, the method may provide useful information on disease progression (monitoring) and detection of biomarkers characterizing early-stage CP.

Correction: Automatic emphysema detection using weakly labeled HRCT lung images.

[This corrects the article DOI: 10.1371/journal.pone.0205397.].

DeepQSM - using deep learning to solve the dipole inversion for quantitative susceptibility mapping.

Quantitative susceptibility mapping (QSM) is based on magnetic resonance imaging (MRI) phase measurements and has gained broad interest because it yields relevant information on biological tissue properties, predominantly myelin, iron and calcium in vivo. Thereby, QSM can also reveal pathological changes of these key components in widespread diseases such as Parkinson's disease, Multiple Sclerosis, or hepatic iron overload. While the ill-posed field-to-source-inversion problem underlying QSM is conventionally assessed by the means of regularization techniques, we trained a fully convolutional deep neural network - DeepQSM - to directly invert the magnetic dipole kernel convolution. DeepQSM learned the physical forward problem using purely synthetic data and is capable of solving the ill-posed field-to-source inversion on in vivo MRI phase data. The magnetic susceptibility maps reconstructed by DeepQSM enable identification of deep brain substructures and provide information on their respective magnetic tissue properties. In summary, DeepQSM can invert the magnetic dipole kernel convolution and delivers robust solutions to this ill-posed problem.

Prostate zonal segmentation in 1.5T and 3T T2W MRI using a convolutional neural network.

Zonal segmentation of the prostate gland using magnetic resonance imaging (MRI) is clinically important for prostate cancer (PCa) diagnosis and image-guided treatments. A two-dimensional convolutional neural network (CNN) based on the U-net architecture was evaluated for segmentation of the central gland (CG) and peripheral zone (PZ) using a dataset of 40 patients (34 PCa positive and 6 PCa negative) scanned on two different MRI scanners (1.5T GE and 3T Siemens). Images were cropped around the prostate gland to exclude surrounding tissues, resampled to 0.5 × 0.5 × 0.5 mm voxels and z -score normalized before being propagated through the CNN. Performance was evaluated using the Dice similarity coefficient (DSC) and mean absolute distance (MAD) in a fivefold cross-validation setup. Overall performance showed DSC of 0.794 and 0.692, and MADs of 3.349 and 2.993 for CG and PZ, respectively. Dividing the gland into apex, mid, and base showed higher DSC for the midgland compared to apex and base for both CG and PZ. We found no significant difference in DSC between the two scanners. A larger dataset, preferably with multivendor scanners, is necessary for validation of the proposed algorithm; however, our results are promising and have clinical potential.

Assessment of prostate cancer prognostic Gleason grade group using zonal-specific features extracted from biparametric MRI using a KNN classifier.

PURPOSE: To automatically assess the aggressiveness of prostate cancer (PCa) lesions using zonal-specific image features extracted from diffusion weighted imaging (DWI) and T2W MRI. METHODS: Region of interest was extracted from DWI (peripheral zone) and T2W MRI (transitional zone and anterior fibromuscular stroma) around the center of 112 PCa lesions from 99 patients. Image histogram and texture features, 38 in total, were used together with a k-nearest neighbor classifier to classify lesions into their respective prognostic Grade Group (GG) (proposed by the International Society of Urological Pathology 2014 consensus conference). A semi-exhaustive feature search was performed (1-6 features in each feature set) and validated using threefold stratified cross validation in a one-versus-rest classification setup. RESULTS: Classifying PCa lesions into GGs resulted in AUC of 0.87, 0.88, 0.96, 0.98, and 0.91 for GG1, GG2, GG1 + 2, GG3, and GG4 + 5 for the peripheral zone, respectively. The results for transitional zone and anterior fibromuscular stroma were AUC of 0.85, 0.89, 0.83, 0.94, and 0.86 for GG1, GG2, GG1 + 2, GG3, and GG4 + 5, respectively. CONCLUSION: This study showed promising results with reasonable AUC values for classification of all GG indicating that zonal-specific imaging features from DWI and T2W MRI can be used to differentiate between PCa lesions of various aggressiveness.

SHARQnet - Sophisticated harmonic artifact reduction in quantitative susceptibility mapping using a deep convolutional neural network.

Quantitative susceptibility mapping (QSM) reveals pathological changes in widespread diseases such as Parkinson's disease, Multiple Sclerosis, or hepatic iron overload. QSM requires multiple processing steps after the acquisition of magnetic resonance imaging (MRI) phase measurements such as unwrapping, background field removal and the solution of an ill-posed field-to-source-inversion. Current techniques utilize iterative optimization procedures to solve the inversion and background field correction, which are computationally expensive and lead to suboptimal or over-regularized solutions requiring a careful choice of parameters that make a clinical application of QSM challenging. We have previously demonstrated that a deep convolutional neural network can invert the magnetic dipole kernel with a very efficient feed forward multiplication not requiring iterative optimization or the choice of regularization parameters. In this work, we extended this approach to remove background fields in QSM. The prototype method, called SHARQnet, was trained on simulated background fields and tested on 3T and 7T brain datasets. We show that SHARQnet outperforms current background field removal procedures and generalizes to a wide range of input data without requiring any parameter adjustments. In summary, we demonstrate that the solution of ill-posed problems in QSM can be achieved by learning the underlying physics causing the artifacts and removing them in an efficient and reliable manner and thereby will help to bring QSM towards clinical applications.

Quality assessment of Ki67 staining using cell line proliferation index and stain intensity features.

Breast cancer is the most frequent cancer among women worldwide. Ki67 can be used as an immunohistochemical pseudo marker for cell proliferation to determine how aggressive the cancer is and thereby the treatment of the patient. No standard Ki67 staining protocol exists, resulting in inter-laboratory stain variability. Therefore, it is important to determine the quality control of a staining protocol to ensure correct diagnosis and treatment of patients. Currently, quality control is performed by the organization NordiQC that use an expert panel-based qualitative assessment system. However, no objective method exists to determine the quality of a staining protocol. In this study, we propose an algorithm, to objectively assess staining quality from segmented cell nuclei structures extracted from cell lines. The cell nuclei were classified into either Ki67 positive or negative to determine the Ki67 proliferation index within the cell lines. A Ki67 stain quality model based on ordinal logistic regression was developed to determine the quality of a staining protocol from features extracted from the segmented cell nuclei in the cell lines. The algorithm was able to segment and classify Ki67 positive cell nuclei with a sensitivity and positive predictive value (PPV) of 0.90 and 0.94 and Ki67 negative cell nuclei with a sensitivity and PPV of 0.78 and 0.78. The mean difference between a manual and automatic Ki67 proliferation index was -0.003 with a standard deviation of 0.056. The ordinal logistic regression model found that the stain intensity for both the Ki67 positive and Ki67 negative cell nuclei were statistically significant as parameters determining the stain quality from the cell line cores. The framework shows great promise for using cell nuclei information from cell lines to predict the staining quality of staining protocols. © 2018 International Society for Advancement of Cytometry.

Automatic emphysema detection using weakly labeled HRCT lung images.

PURPOSE: A method for automatically quantifying emphysema regions using High-Resolution Computed Tomography (HRCT) scans of patients with chronic obstructive pulmonary disease (COPD) that does not require manually annotated scans for training is presented. METHODS: HRCT scans of controls and of COPD patients with diverse disease severity are acquired at two different centers. Textural features from co-occurrence matrices and Gaussian filter banks are used to characterize the lung parenchyma in the scans. Two robust versions of multiple instance learning (MIL) classifiers that can handle weakly labeled data, miSVM and MILES, are investigated. Weak labels give information relative to the emphysema without indicating the location of the lesions. The classifiers are trained with the weak labels extracted from the forced expiratory volume in one minute (FEV1) and diffusing capacity of the lungs for carbon monoxide (DLCO). At test time, the classifiers output a patient label indicating overall COPD diagnosis and local labels indicating the presence of emphysema. The classifier performance is compared with manual annotations made by two radiologists, a classical density based method, and pulmonary function tests (PFTs). RESULTS: The miSVM classifier performed better than MILES on both patient and emphysema classification. The classifier has a stronger correlation with PFT than the density based method, the percentage of emphysema in the intersection of annotations from both radiologists, and the percentage of emphysema annotated by one of the radiologists. The correlation between the classifier and the PFT is only outperformed by the second radiologist. CONCLUSIONS: The presented method uses MIL classifiers to automatically identify emphysema regions in HRCT scans. Furthermore, this approach has been demonstrated to correlate better with DLCO than a classical density based method or a radiologist, which is known to be affected in emphysema. Therefore, it is relevant to facilitate assessment of emphysema and to reduce inter-observer variability.

Motion analysis of the cervical spine during extension and flexion: Reliability of the vertebral marking procedure.

Cervical spine motion analysis using videofluoroscopy is currently a technique without a gold standard. We demonstrate the reliability of a rigid and reliable analysis methodology for cervical motion using videofluoroscopic images, representing the entire range of motion during flexion and extension, from the neutral position to the end-range in the sagittal plane. Two researchers with radiography and vertebral marking expertise, and two inexperienced researchers with 10 hours of training manually marked anatomical structures on fluoroscopic images in a procedure designed to control for vertebral rotation around the mid-plane axis. The average marking error across examiners and images was -0.12∘ (standard deviation: 0.88°), and the intraexaminer error ranged from -1.00∘ to 1.61° (standard deviation range: 0.27°-1.19°). Our method demonstrated lower errors compared to the higher resolution X-ray studies, and proved that vertebral marking can be performed by persons with no experience in radiographic image analysis.

Postpartum placental CT angiography in normal pregnancies and in those complicated by diabetes mellitus.

INTRODUCTION: Pregnancy complicated by diabetes mellitus (DM) is a central obstetric problem often complicated by fetal macrosomia and increased risk of intrapartum asphyxia. This risk might be explained by fetoplacental vascular abnormalities. This study aimed to investigate the fetoplacental vascular volume by placental CT angiography in normal pregnancies and in pregnancies complicated by type 1 DM (T1DM), diet controlled gestational DM (GDMd), and insulin treated gestational DM (GDMi). METHODS: Postpartum, barium contrast enhanced placental CT angiography was performed in 27 normal pregnancies and 25 DM pregnancies (8 T1DM, 8 GDMd, and 9 GDMi). The fetoplacental vascular volume/placenta weight (FVV/PW)-ratio and fetoplacental vascular volume/birth weight (FVV/BW)-ratio of each diabetic group were compared to the normal group with multiple regression analysis adjusted for GA. In all pregnancies a standardized histopathological placental examination was performed postpartum. RESULTS: In normal pregnancies, the fetoplacental vascular volume increased with GA (p < 0.001), placental weight (p < 0.001), and birth weight (p < 0.001). In T1DM and GDMi pregnancies, the gestational age adjusted placental weight and the birth weight were increased when compared to normal pregnancies (p < 0.05). The FVV/BW-ratio was significantly reduced in both T1DM and GDMi pregnancies when compared to normal pregnancies (p = 0.003 and p = 0.009, respectively). DISCUSSION: This study demonstrates, that in insulin treated DM pregnancies the fetus as well as the placenta is larger than normal. However, despite a large placenta, a relatively smaller fetoplacental vascular volume supplies the macrosomic fetus. This finding might explain why fetuses from insulin treated DM pregnancies have high vulnerability to intrauterine and intrapartum asphyxia.

Repeatability of Cervical Joint Flexion and Extension Within and Between Days.

OBJECTIVE: The purpose of this study was to investigate within- and between-day repeatability of free and unrestricted healthy cervical flexion and extension motion when assessing dynamic cervical spine motion. METHODS: Fluoroscopy videos of 2 repeated cervical flexion and 2 repeated extension motions were examined for within-day repeatability (20-second interval) for 18 participants (6 females) and between-day repeatability (1-week interval) for 15 participants (6 females). The dynamic cervical motions were free and unrestricted from neutral to end range. The flexion videos and extension videos were evenly divided into 10% epochs of the C0-to-C7 range of motion. Within-day and between-day repeatability of joint motion angles (all 7 joints and epochs, respectively) was tested in a repeated-measures analysis of variance. Joint motion angle differences between repetitions were calculated for each epoch and joint (7 joints), and these joint motion angle differences between within-day and between-day repetitions were tested in mixed-model analysis of variance. RESULTS: For all joints and epochs, respectively, no significant differences were found in joint motion angle between within-day or between-day repetitions. There were no significant effects of joint motion angle differences between within-day and between-day repetitions. The average within-day joint motion angle differences across all joints and epochs were 0.00° ± 2.98° and 0.00° ± 3.05° for flexion and extension, respectively. The average between-day joint motion angle differences were 0.02° ± 2.56° and 0.05° ± 2.40° for flexion and extension, respectively. CONCLUSIONS: This is the first study to report the within-day and between-day joint motion angle differences of repeated cervical flexion and extension. This study supports the idea that cervical joints repeat their motion accurately.

Accurate Measurement of Airway Morphology on Chest CT Images.

In recent years, the ability to accurately measuring and analyzing the morphology of small pulmonary structures on chest CT images, such as airways, is becoming of great interest in the scientific community. As an example, in COPD the smaller conducting airways are the primary site of increased resistance in COPD, while small changes in airway segments can identify early stages of bronchiectasis. To date, different methods have been proposed to measure airway wall thickness and airway lumen, but traditional algorithms are often limited due to resolution and artifacts in the CT image. In this work, we propose a Convolutional Neural Regressor (CNR) to perform cross-sectional measurements of airways, considering wall thickness and airway lumen at once. To train the networks, we developed a generative synthetic model of airways that we refined using a Simulated and Unsupervised Generative Adversarial Network (SimGAN). We evaluated the proposed method by first computing the relative error on a dataset of synthetic images refined with SimGAN, in comparison with other methods. Then, due to the high complexity to create an in-vivo ground-truth, we performed a validation on an airway phantom constructed to have airways of different sizes. Finally, we carried out an indirect validation analyzing the correlation between the percentage of the predicted forced expiratory volume in one second (FEV1%) and the value of the Pi10 parameter. As shown by the results, the proposed approach paves the way for the use of CNNs to precisely and accurately measure small lung airways with high accuracy.

Using cell nuclei features to detect colon cancer tissue in hematoxylin and eosin stained slides.

Currently, diagnosis of colon cancer is based on manual examination of histopathological images by a pathologist. This can be time consuming and interpretation of the images is subject to inter- and intra-observer variability. This may be improved by introducing a computer-aided diagnosis (CAD) system for automatic detection of cancer tissue within whole slide hematoxylin and eosin (H&E) stains. Cancer disrupts the normal control mechanisms of cell proliferation and differentiation, affecting the structure and appearance of the cells. Therefore, extracting features from segmented cell nuclei structures may provide useful information to detect cancer tissue. A framework for automatic classification of regions of interest (ROI) containing either benign or cancerous colon tissue extracted from whole slide H&E stained images using cell nuclei features was proposed. A total of 1,596 ROI's were extracted from 87 whole slide H&E stains (44 benign and 43 cancer). A cell nuclei segmentation algorithm consisting of color deconvolution, k-means clustering, local adaptive thresholding, and cell separation was performed within the ROI's to extract cell nuclei features. From the segmented cell nuclei structures a total of 750 texture and intensity-based features were extracted for classification of the ROI's. The nine most discriminative cell nuclei features were used in a random forest classifier to determine if the ROI's contained benign or cancer tissue. The ROI classification obtained an area under the curve (AUC) of 0.96, sensitivity of 0.88, specificity of 0.92, and accuracy of 0.91 using an optimized threshold. The developed framework showed promising results in using cell nuclei features to classify ROIs into containing benign or cancer tissue in H&E stained tissue samples. © 2017 International Society for Advancement of Cytometry.

An Improved Algorithm for Coronary Bypass Anastomosis Segmentation in Epicardial Ultrasound Sequences.

Epicardial ultrasound (EUS) can be used for intra-operative quality assessment of coronary artery bypass anastomoses. To quantify the anastomotic quality from EUS images, the area of anastomotic structures has to be extracted from EUS sequences. Currently, this is done manually as no objective methods are available. We used an automatic anastomosis segmentation algorithm to extract the area of anastomotic structures from in vivo EUS sequences obtained from 16 porcine anastomoses. The algorithm consists of four major components: vessel detection, vessel segmentation, segmentation quality control and inter-frame contour alignment. The segmentation accuracy was assessed using m-fold cross-validation based on 830 manual segmentations of the anastomotic structures. A Dice coefficient of 0.879 (±0.073) and an absolute area difference of 16.95% (±17.94) were obtained. The proposed segmentation algorithm has potential to automatically extract the area of anastomotic structures.