Semi-automatic Method Research Articles

Kinect v2-Assisted Semi-Automated Method to Assess Upper Limb Motor Performance in Children.

The interruption of rehabilitation activities caused by the COVID-19 lockdown has significant health negative consequences for the population with physical disabilities. Thus, measuring the range of motion (ROM) using remotely taken photographs, which are then sent to specialists for formal assessment, has been recommended. Currently, low-cost Kinect motion capture sensors with a natural user interface are the most feasible implementations for upper limb motion analysis. An active range of motion (AROM) measuring system based on a Kinect v2 sensor for upper limb motion analysis using Fugl-Meyer Assessment (FMA) scoring is described in this paper. Two test groups of children, each having eighteen participants, were analyzed in the experimental stage, where upper limbs’ AROM and motor performance were assessed using FMA. Participants in the control group (mean age of 7.83 ± 2.54 years) had no cognitive impairment or upper limb musculoskeletal problems. The study test group comprised children aged 8.28 ± 2.32 years with spastic hemiparesis. A total of 30 samples of elbow flexion and 30 samples of shoulder abduction of both limbs for each participant were analyzed using the Kinect v2 sensor at 30 Hz. In both upper limbs, no significant differences (p < 0.05) in the measured angles and FMA assessments were observed between those obtained using the described Kinect v2-based system and those obtained directly using a universal goniometer. The measurement error achieved by the proposed system was less than ±1° compared to the specialist’s measurements. According to the obtained results, the developed measuring system is a good alternative and an effective tool for FMA assessment of AROM and motor performance of upper limbs, while avoiding direct contact in both healthy children and children with spastic hemiparesis.

Segmentation of acetowhite region in uterine cervical image based on deep learning

Acetowhite (AW) region is a critical physiological phenomenon of precancerous lesions of cervical cancer. An accurate segmentation of the AW region can provide a useful diagnostic tool for gynecologic oncologists in screening cervical cancers. Traditional approaches for the segmentation of AW regions relied heavily on manual or semi-automatic methods. To automatically segment the AW regions from colposcope images. First, the cervical region was extracted from the original colposcope images by k-means clustering algorithm. Second, a deep learning-based image semantic segmentation model named DeepLab V3+ was used to segment the AW region from the cervical image. The results showed that, compared to the fuzzy clustering segmentation algorithm and the level set segmentation algorithm, the new method proposed in this study achieved a mean Jaccard Index (JI) accuracy of 63.6% (improved by 27.9% and 27.5% respectively), a mean specificity of 94.9% (improved by 55.8% and 32.3% respectively) and a mean accuracy of 91.2% (improved by 38.6% and 26.4% respectively). A mean sensitivity of 78.2% was achieved by the proposed method, which was 17.4% and 10.1% lower respectively. Compared to the image semantic segmentation models U-Net and PSPNet, the proposed method yielded a higher mean JI accuracy, mean sensitivity and mean accuracy. The improved segmentation performance suggested that the proposed method may serve as a useful complimentary tool in screening cervical cancer.

SEEMLIS: a flexible semi-automated method for enrichment of methylated DNA from low-input samples

BackgroundDNA methylation alterations have emerged as hallmarks of cancer and have been proposed as screening, prognostic, and predictive biomarkers. Traditional approaches for methylation analysis have relied on bisulfite conversion of DNA, which can damage DNA and is not suitable for targeted gene analysis in low-input samples. Here, we have adapted methyl-CpG-binding domain protein 2 (MBD2)-based DNA enrichment for use on a semi-automated exclusion-based sample preparation (ESP) platform for robust and scalable enrichment of methylated DNA from low-input samples, called SEEMLIS.ResultsWe show that combining methylation-sensitive enzyme digestion with ESP-based MBD2 enrichment allows for single gene analysis with high sensitivity for GSTP1 in highly impure, heterogenous samples. We also show that ESP-based MBD2 enrichment coupled with targeted pre-amplification allows for analysis of multiple genes with sensitivities approaching the single cell level in pure samples for GSTP1 and RASSF1 and sensitivity down to 14 cells for these genes in highly impure samples. Finally, we demonstrate the potential clinical utility of SEEMLIS by successful detection of methylated gene signatures in circulating tumor cells (CTCs) from patients with prostate cancer with varying CTC number and sample purity.ConclusionsSEEMLIS is a robust assay for targeted DNA methylation analysis in low-input samples, with flexibility at multiple steps. We demonstrate the feasibility of this assay to analyze DNA methylation in prostate cancer cells using CTCs from patients with prostate cancer as a real-world example of a low-input analyte of clinical importance. In summary, this novel assay provides a platform for determining methylation signatures in rare cell populations with broad implications for research as well as clinical applications.

Can preoperative features predict residual false lumen remodelling after tear-oriented limited resection for acute type I dissection?

Our goal was to identify the preoperative findings in computed tomography correlated with the postoperative changes of the false lumen (FL) remaining in the descending thoracic aorta following tear-oriented surgery for acute type I dissection. Patients who underwent ascending +/- partial arch replacement, with available preoperative and postoperative late (>1 year) CT scans, were included. Preoperative cross-sectional parameters were measured by the semi-automated centreline method at the level of the anastomosis. The parameters of the patients who presented positive remodelling of the proximal descending thoracic aorta were compared with those of the patients who did not in the late images. Among the included 101 patients, positive remodelling of the proximal descending thoracic aorta was observed in 46.5%, of which 76.6% extended downwards to the middle descending thoracic aorta. In the univariable analysis, an FL area ratio <50% (P < 0.001), a circumferential ratio of dissection <50% (P = 0.028), an FL width <20 mm (P = 0.008) at the distal anastomotic zone and not leaving residual arch branches having patent false lumens (P = 0.005) correlated with positive remodelling. The number of fulfilled above-mentioned features revealed a better correlation, which was stronger in patients without Marfan syndrome and in those older than 50 years. The cross-sectional extent of dissection at the presumed distal anastomotic zone is associated with descending thoracic aorta positive remodelling following tear-oriented replacement for acute type I aortic dissection. Considering the anatomical features in determining the extent of aortic replacement, some of the non-Marfan elderly patients can be spared from aggressive total arch replacement with the frozen elephant trunk technique.

A Semi-Automated Method for Measuring Biofilm Accumulation on the Teeth Using Quantitative Light-Induced Fluorescence in Dogs and Cats.

Oral health conditions (eg, plaque, calculus, gingivitis) cause morbidity and pain in companion animals. Thus, developing technologies that can ameliorate the accumulation of oral biofilm, a critical factor in the progression of these conditions, is vital. Quantitative light-induced fluorescence (QLF) is a method to quantify oral substrate accumulation, and therefore, it can assess biofilm attenuation of different products. New software has recently been developed that automates aspects of the procedure. However, few QLF studies in companion animals have been performed. QLF was used to collect digital images of oral substrate accumulation on the teeth of dogs and cats to demonstrate the ability of QLF to discriminate between foods known to differentially inhibit oral substrate accumulation. Images were taken as a function of time and diet. Software developed by the Cytometry Laboratory, Purdue University quantified biofilm coverage. Intra- and intergrader reproducibility was also assessed, as was a comparison of the results of the QLF software with those of an experienced grader using undisclosed coverage-only metrics similar to those used for the Logan and Boyce index. Quantification of oral substrate accumulation using QLF-derived images demonstrated the ability to distinguish between dental diets known to differentially inhibit oral biofilm accumulation. Little variance in intra- and intergrader reproducibility was observed, and the comparison between the experienced Logan and Boyce grader and the QLF software yielded a concordance correlation coefficient of 0.89 (95% CI = 0.84, 0.92). These results show that QLF is a useful tool that allows the semi-automated quantification of the accumulation of oral biofilm in companion animals.

Development of a vortex oscillating clean-up column for high-throughput semi-automatic sample preparation of drug residues in fish muscle tissues

A novelty high-throughput semi-automatic sample clean-up method was developed and used for rapid analysis of 9 drug residues in fish muscle tissues combined with high-performance liquid chromatography/quadrupole-linear ion trap mass spectrometry (HPLC-QTRAP-MS/MS). The target compounds were extracted by Quick, Easy, Cheap, Effective, Rugged and Safe multiresidue (QuEChERS) method with acetonitrile. A vortex oscillating clean-up column was proposed for single step purification firstly. After centrifugation, 1 mL of acetonitrile layer was transferred into a 3 mL vortex oscillating clean-up column which packed with 150 mg C18 sorbent. The column was covered with piston, placed on a column-tube rack, vigorously agitated on a multi-tube vortex mixer. After vortex oscillation, purified extract was obtained for analysis by pushing the piston of oscillating clean-up column. The performance of the developed method was validated by analyzing 9 analytes in spiked grass carp muscle tissues. The limits of quantification (LOQs) ranged from 1.0 to 5.0 μg kg−1. The recoveries were in the range of 101–119% at the four concentration levels (2.0, 5.0, 10.0, 20.0 μg kg−1). Inter-day and intra-day relative standard deviations (RSDs) were lower than 16% for all analytes.

Relationship between epicardial fat volume on cardiac CT and atherosclerosis severity in three-vessel coronary artery disease: a single-center cross-sectional study

BackgroundThe ideal treatment strategy for stable three-vessel coronary artery disease (CAD) patients are difficult to determine and for patients undergoing conservative treatment, imaging evidence of coronary atherosclerotic severity progression remains limited. Epicardial fat volume (EFV) on coronary CT angiography (CCTA) has been considered to be associated with coronary atherosclerosis. Therefore, this study aims to evaluate the relationship between EFV level and coronary atherosclerosis severity in three-vessel CAD.MethodsThis retrospective study enrolled 252 consecutive patients with three-vessel CAD and 252 normal control group participants who underwent CCTA between January 2018 and December 2019. A semi-automatic method was developed for EFV quantification on CCTA images, standardized by body surface area. Coronary atherosclerosis severity was evaluated and scored by the number of coronary arteries with ≥ 50% stenosis on coronary angiography. Patients were subdivided into groups on the basis of lesion severity: mild (score = 3 vessels, n = 85), moderate (3.5 vessels ≤ score < 4 vessels, n = 82), and severe (4 vessels ≤ score ≤ 7 vessels, n = 85). The independent sample t-test, analysis of variance, and logistic regression analysis were used to evaluate the associations between EFV level and severity of coronary atherosclerosis.ResultsCompared with normal controls, three-vessel CAD patients had significantly higher EFV level (65 ± 22 mL/m2 vs. 48 ± 19 mL/m2; P < 0.001). In patients with three-vessel CAD, there was a progressive decline in EFV level as the score of coronary atherosclerosis severity increased, especially in those patients with a body mass index (BMI) ≥ 25 kg/m2 (75 ± 21 mL/m2 vs. 72 ± 22 mL/m2 vs. 62 ± 17 mL/m2; P < 0.05). Multivariable regression analysis showed that both BMI (OR 3.40, 95% CI 2.00–5.78, P < 0.001) and the score of coronary atherosclerosis severity (OR 0.49, 95% CI 0.26–0.93, P < 0.05) were independently related to the change of EFV level.ConclusionThree-vessel CAD patients do have higher EFV level than the normal controls. While, there may be an inverse relationship between EFV level and the severity of coronary atherosclerosis in patients with three-vessel CAD.

Cone-Beam Computed Tomography-Based Spatial Prediction of Drug Dose After Transarterial Chemoembolization Using Radiopaque Drug-Eluting Beads in Woodchuck Hepatocellular Carcinoma.

The aims of this study were to develop a model to estimate drug dose delivered to tumors after transarterial chemoembolization (TACE) with radiopaque drug-eluting beads (DEBs) based on DEB density on cone-beam computed tomography (CT) and to evaluate drug penetration into tissue in a woodchuck hepatoma model. Transarterial chemoembolization was performed in woodchucks with hepatocellular carcinoma (N = 5) using DEBs (70-150 μm, LC Bead LUMI) loaded with doxorubicin. Livers were resected 45 minutes after embolization, immediately frozen, and cut using liver-specific, 3D-printed sectioning molds. Doxorubicin levels in tumor specimens were measured by high-performance liquid chromatography and correlated with DEB iodine content that was measured using prototype cone-beam CT-based embolization treatment planning software. Doxorubicin penetration into tissue surrounding DEBs was assessed by fluorescence microscopy of tumor sections. Fluorescence intensity was converted into doxorubicin concentration using calibration standards. Intensity-thresholded color heatmaps were generated representing extravascular drug penetration. Consistent segmentation of DEBs on cone-beam CT was achieved using a semiautomated intensity thresholding method. A positive linear correlation (0.96) was found between DEB iodine content measured on cone-beam CT and the amount of doxorubicin measured in tumor specimens. Prediction of doxorubicin levels in tumor sections that were not included in model development was accurate, with a root-mean-square error of 0.08 mg of doxorubicin. Tumor penetration of eluted doxorubicin resulted in concentration gradients where drug content decreased with increasing distance from blood vessels containing DEBs. Drug penetration was greater for blood vessels containing DEB clusters compared with single DEB, with higher doxorubicin concentrations extending further away from the vessels. Estimation of drug dose delivered during transarterial chemoembolization in a woodchuck hepatocellular carcinoma model was possible using DEB radiopacity on cone-beam CT as a surrogate marker. Doxorubicin penetration was greatest adjacent to vessels containing DEB clusters compared with single DEB. Intraprocedural estimation of the spatial distribution of drug dose within the tumor could enable real-time adjustments to DEB delivery, to maximize treatment coverage or identify regions of tumor at risk for undertreatment.

Semi-Automated Training of Rat Ultrasonic Vocalizations.

Rats produce ultrasonic vocalizations (USVs) for conspecific communication. These USVs are valuable biomarkers for studying behavioral and mechanistic changes in a variety of diseases and disorders. Previous work has demonstrated operant conditioning can progressively increase the number of USVs produced by rats over multiple weeks. This operant conditioning paradigm is a useful model for investigating the effects of increased laryngeal muscle use on USV acoustic characteristics and underlying central and peripheral laryngeal sensorimotor mechanisms. Previous USV operant conditioning studies relied on manual training to elicit USV productions, which is both time and labor intensive and can introduce human variability. This manuscript introduces a semi-automated method for training rats to increase their rate of USV production by pairing commercially available operant conditioning equipment with an ultrasonic detection system. USV training requires three basic components: elicitation cue, detection of the behavior, and a reward to reinforce the desired behavior. With the semi-automated training paradigm, indirect exposure to the opposite sex or an olfactory cue can be used to elicit USV production. The elicited USV is then automatically detected by the ultrasonic acoustic system, which consequently triggers the release of a sucrose pellet reward. Our results demonstrate this semi-automated procedure produces a similar increase in USV production as the manual training method. Through automation of USV detection and reward administration, staffing requirements, human error, and subject behavioral variability may be minimized while scalability and reproducibility are increased. This automation may also result in greater experimental flexibility, allowing USV training paradigms to become more customizable for a wider array of applications. This semi-automated USV behavioral training paradigm improves upon manual training techniques by increasing the ease, speed, and quality of data collection.

Characterization of Macrophage-Like Cells in Retinal Vein Occlusion Using En Face Optical Coherence Tomography.

PurposeTo investigate the clinical features of a macrophage-like cell (MLC) obtained by en face optical coherence tomography (OCT) in retinal vein occlusion (RVO).MethodsThe study involved 36 patients with treatment-naïve unilateral acute RVO, including 21 branch RVO (BRVO) and 15 central RVO. Vessel density and macular thickness were quantified using OCT angiography. A 3-μm en face OCT slab on the inner limiting membrane in the optic nerve head (ONH) region or macular region was used to visualize the MLCs. The MLCs were binarized and quantified using a semiautomated method. The unaffected fellow eyes served as the control group.ResultsThe morphology of MLCs appeared larger and plumper in RVO eyes. The mean MLC density in the ONH and macular regions was 2.46 times and 2.86 times higher than their fellow eyes, respectively (p < 0.001). The macular MLC density of the occlusive region was significantly lower than that of the unaffected region in BRVO (p = 0.01). The ONH and macular MLC densities in the non-perfused region were significantly lower than those in the perfused region in all RVO eyes (p < 0.001). The ONH MLC density in RVO eyes was negatively correlated with radial peripapillary capillary vessel density (r = −0.413, p = 0.012). Both ONH and macular MLC densities were positively correlated with macular thickness (r = 0.505, p = 0.002; r = 0.385, p = 0.02, respectively).ConclusionThe increased density and changes of morphology characterized by OCT may indicate generalized activation and aggregation of MLCs in RVO. More MLCs are recruited in the perfused region rather than the non-perfused region. RVO eyes with a higher density of MLCs tend to suffer from the thicker macula.

Isotropic 3D compressed sensing (CS) based sequence is comparable to 2D-LGE in left ventricular scar quantification in different disease entities

The goal of this study was to evaluate a three-dimensional compressed sensing (3D-CS) LGE prototype sequence for the detection and quantification of myocardial fibrosis in patients with chronic myocardial infarction (CMI) and myocarditis (MYC) compared with a 2D-LGE standard. Patients with left-ventricular LGE due to CMI (n = 33) or MYC (n = 20) were prospectively recruited. 2D-LGE and 3D-CS images were acquired in random order at 1.5 Tesla. 3D-CS short axis (SAX) images were reconstructed corresponding to 2D SAX images. LGE was quantitatively assessed on patient and segment level using semi-automated threshold methods. Image quality (4-point scoring system), Contrast-ratio (CR) and acquisition times were compared. There was no significant difference between 2D and 3D sequences regarding global LGE (%) (CMI [2D-LGE: 11.4 ± 7.5; 3D-LGE: 11.5 ± 8.5; p = 0.99]; MYC [2D-LGE: 27.0 ± 15.7; 3D-LGE: 26.2 ± 13.1; p = 0.70]) and segmental LGE-extent (p = 0.63). 3D-CS identified papillary infarction in 5 cases which was not present in 2D images. 2D-LGE acquisition time was shorter (2D: median: 06:59 min [IQR: 05:51–08:18]; 3D: 14:48 min [12:45–16:57]). 3D-CS obtained better quality scores (2D: 2.06 ± 0.56 vs. 3D: 2.29 ± 0.61). CR did not differ (p = 0.63) between basal and apical regions in 3D-CS images but decreased significantly in 2D apical images (CR basal: 2D: 0.77 ± 0.11, 3D: 0.59 ± 0.10; CR apical: 2D: 0.64 ± 0.17, 3D: 0.53 ± 0.11). 3D-LGE shows high congruency with standard LGE and allows better identification of small lesions. However, the current 3D-CS LGE sequence did not provide PSIR reconstruction and acquisition time was longer.

A Novel and Practical Protocol for Three-Dimensional Assessment of Alveolar Cleft Grafting Procedures.

To evaluate the reproducibility and accuracy of a new, easy-to-use volumetric assessment of the alveolar cleft. Twelve cone-beam computed tomography (CBCT) datasets of patients with a unilateral cleft lip, alveolus, and palate were evaluated by two investigators. Residual alveolar cleft calcified volume one year after surgery was analyzed by using standardized landmarks to determine the borders of the cleft defect and semi-automatically segment the alveolar cleft defect. The Dice-coefficient between observers for the segmented preoperative alveolar cleft defect was 0.81. Average percentage of residual alveolar cleft calcified material was 66.7% one year postoperatively. This study demonstrates a reliable and practical semi-automatic three-dimensional volumetric assessment method for unilateral clefts using CBCT.

Zika virus congenital microcephaly severity classification and the association of severity with neuropsychomotor development.

Zika virus infection during pregnancy is linked to birth defects, most notably microcephaly, which is associated with neurodevelopmental delays. The goals of the study were to propose a method for severity classification of congenital microcephaly based on neuroradiologic findings of MRI scans, and to investigate the association of severity with neuropsychomotor developmental scores. We also propose a semi-automated method for MRI-based severity classification of microcephaly. We conducted a cross-sectional investigation of 42 infants born with congenital Zika infection. Bayley Scales of Infant and Toddler Development III (Bayley-III) developmental evaluations and MRI scans were carried out at ages 13-39months (mean: 24.8months; standard deviation [SD]: 5.8months). The severity score was generated based on neuroradiologist evaluations of brain malformations. Next, we established a distribution of Zika virus-microcephaly severity score including mild, moderate and severe and investigated the association of severity with neuropsychomotor developmental scores. Finally, we propose a simplified semi-automated procedure for estimating the severity score based only on volumetric measures. The results showed a correlation of r=0.89 (P<0.001) between the Zika virus-microcephaly severity score and the semi-automated method. The trimester of infection did not correlate with the semi-automated method. Neuropsychomotor development correlated with the severity classification based on the radiologic readings and semi-automated method; the more severe the imaging scores, the lower the neuropsychomotor developmental scores. These severity classification methods can be used to evaluate severity of microcephaly and possible association with developmental consequences. The semi-automated methods thus provide an alternative for predicting severity of microcephaly based on only one MRI sequence.

Trajectory analysis at mid-block section using microscopic simulation

With rapid urbanization, traffic has increased in recent years. For a developing nation, mobility is a key concern. Especially, in rapidly expanding urban areas, vehicle accidents are of great concern. Because of this, it is necessary to pay vital attention to transportation and examine the need for greater geometric design and capacity. To assess road user interactions at a mid-block with heterogeneous traffic complexity, innovative trajectory-based data was used. In order to evaluate microscopic traffic flow parameters under various traffic flow conditions a Support Vector Machine (SVM) is used to classify severity grades based on specified indicators which are between generated at two mid-block sections. A vehicle trajectory data for two different mid-block sections using a semi-automated image processing method is generated. The trajectories detected are vehicle ID, vehicle time, speed, longitude, latitude position, flow. From the analysis, microscopic traffic flow parameter like time headway, space headway is estimated by plotting a graph using Linear Regression model. This means that SVM is the best fit model to estimate headway at mid-block section (R2 = 0.86, 0.85) studied. From this study, it is concluded that the obtained value of R2 and goodness of fit measures using trajectory data base is highly acceptable to estimate the headway at mid-block section.

Applying ontology learning and multi-objective ant colony optimization method for focused crawling to meteorological disasters domain knowledge

The focused crawler based on semantic analysis is a research hotspot in the field of information retrieval. The domain ontology is generally applied to construct the topic model of the focused crawler. In order to overcome the limitations of builders' knowledge reserve and subjective consciousness in the process of constructing artificially ontology, a semi-automatic construction method of domain ontology based on ontology learning technology combining the latent Dirichlet allocation and the Apriori algorithm is proposed in this article. When evaluating the relevance between a hyperlink and a specific topic, the joint evaluation method considering both the web text and the link structure is usually used. However, the traditional weighted sum method is difficult to reasonably determine the optimal weights of these evaluating indicators. To solve this problem, a multi-objective optimization model for link evaluation and a subsequent multi-objective ant colony optimization algorithm (MOACO) are proposed. In the MOACO, a method of the nearest farthest candidate solution (NFCS) is combined with the fast non-dominated sorting to select a set of Pareto-optimal hyperlinks and guide the crawlers’ search directions. The experimental results of the focused crawling on the domain knowledge of typhoon disasters and rainstorm disasters prove that the ability of the proposed focused crawlers to retrieve topic-relevant webpages.

Distortion Calculation Method Based on Image Processing for Automobile Lateral Mirrors.

The automobile lateral-view mirrors are the most important visual support for driver safety; therefore, it is important they have robust quality control. Typically, the distortion of a lateral-view mirror is measured using the JIS-D-5705 standard; however, this methodology requires an expert person to perform the measurements and calculations manually, which can induce measurement errors. In this work, a semi-automatic distortion calculation method based on image processing is presented. Distortion calculations of five commercial mirrors from different manufacturers were performed, and a comparative study was carried out between the JIS-D-5705 standard and the proposed method. Experimental results performed according to the JIS-D-5705 standard showed that all mirrors have a distortion lower than 5%, indicating that all meet the standard. On the other hand, the proposed method was able to detect that one of the mirrors presented an important distortion, which was not detected by the methodology proposed in the standard; therefore, that mirror should not meet the standard. Then, it was possible to conclude that the proposed distortion calculation method, based on image processing, has higher robustness and precision than the standard. In addition, an appropriate and effective behavior against changes in scale, resolution, and, unlike the standard, against changes in image rotation was also shown.

Convolutional Neural Network to Segment Laminae on 3D Ultrasound Spinal Images to Assist Cobb Angle Measurement.

A recent innovation in scoliosis monitoring is the use of ultrasonography, which provides true 3D information in one scan and does not emit ionizing radiation. Measuring the severity of scoliosis on ultrasonographs requires identifying lamina pairs on the most tilted vertebrae, which is difficult and time-consuming. To expedite and automate measurement steps, this paper detailed an automatic convolutional neural network-based algorithm for identifying the laminae on 3D ultrasonographs. The predicted laminae were manually paired to measure the lateral spinal curvature on the coronal view, called the Cobb angle. In total, 130 spinal ultrasonographs of adolescents with idiopathic scoliosis recruited from a scoliosis clinic were selected, with 70 for training and 60 for testing. Data augmentation increased the effective training set size to 140 ultrasonographs. Semi-automatic Cobb measurements were compared to manual measurements on the same ultrasonographs. The semi-automatic measurements demonstrated good inter-method reliability (ICC3,1 = 0.87) and performed better on thoracic (ICC3,1 = 0.91) than lumbar curves (ICC3,1 = 0.81). The mean absolute difference and standard deviation between semi-automatic and manual was 3.6°±3.0°. In conclusion, the semi-automatic method to measure the Cobb angle on ultrasonographs is feasible and accurate. This is the first algorithm that automates steps of Cobb angle measurement on ultrasonographs.

Aggravation of retinal hard exudates after intravitreal anti-vascular endothelial growth factor therapy for cystoid macular edema and the risk factors: a retrospective study

Background/aimsTo evaluate retinal hard exudates (HEs) progression in patients with cystoid macular edema (CME) secondary to diabetic retinopathy (DR) or branch retinal vascular occlusion (BRVO) after intravitreal injections of ranibizumab (IVR) treatment and identify the risk factors for the deterioration of HEs.MethodsThis retrospective study enrolled 288 eyes with center-involving CME secondary to DR or BRVO from 288 patients (one eye per patient). All patients were treated with three loading doses of ranibizumab intravitreally at a monthly interval. The morphologic features of HEs were observed, and the HEs areas were quantified using a semi-automatic method at baseline, 1 month after the first dose of IVR and 1 month after the third dose of IVR therapy. HEs progression was defined as having a > =2-grade increase in the HEs severity scale. The best-corrected vision acuity (BCVA) and alterations in HEs areas were compared between DR and BRVO groups. And logistic regression analyses were used to identify the risk factors for HEs exacerbation.ResultsMorphological changes of retinal HEs occurred in all eyes after IVR therapy, although HEs area was not significantly changed in some eyes. DR group has a higher percentage of eyes with progressed HEs area than the BRVO groups (34.9% vs. 21.8%, P = 0.019) 1 month after the first dose of IVR. Both DR and BRVO groups had a decreased percentage of enlarged HEs 1 month after the third injection, but the DR group is still higher than the BRVO group (17.1% vs. 8.4%, P = 0.027). At baseline, there was no correlation between VA and HEs areas. After the first and third doses of IVR, there still was no consistent correlation between HEs severity and change in VA over time. Furthermore, CME with subretinal fluid (SRF) is associated with a higher risk of HEs progression (P = 0.001). Long CME duration and high serum low-density lipoprotein cholesterol (LDL-C) level were identified as risk factors for HEs progression following IVR treatment in both univariable and multivariable regression analyses (Odds ratio (OR) = 1.88, P = 0.012 and OR = 1.14, P = 0.021, respectively).ConclusionsAlterations in the area of retinal HEs are widely observed after IVR treatment for CME. The eyes with CME secondary to DR have a higher percentage of progressed HEs than the BRVO eyes. DME with SRF, extended duration of CME, and high LDL-C level are potential risk factors of deteriorated HEs after IVR treatment.

Uncovering Vegetation Changes in the Urban–Rural Interface through Semi-Automatic Methods

Forest fires are considered by Portuguese civil protection as one of the most serious natural disasters due to their frequency and extent. To address the problem, the Fire Forest Defense System establishes the implementation of fuel management bands to aid firefighting. The aim of this study was to develop a model capable of identifying vegetation removal in the urban–rural interface defined by law for fuel management actions. The model uses normalised difference vegetation index (NDVI) of Sentinel-2 images time series and is based on the Welch t-test to find statistically significant differences between (i) the value of the NDVI in the pixel; (ii) the mean of the NDVI in the pixels of the same land cover type in a radius of 500 m; and (iii) their difference. The model identifies a change when the t-test points for a significant difference of the NDVI value in the ‘pixel’ as comparted to the ‘difference’ but not the ‘mean’. We use a moving window limited to 60 days before and after the analysed date to reduce the phenological variations of vegetation. The model was applied in five municipalities of Portugal and the results are promising to identify the places where the management of fuel bands was not carried out. This indicates which model could be used to assist in the verification of the annual management of the fuel bands defined in the law.

A multi–modal unsupervised fault detection system based on power signals and thermal imaging via deep AutoEncoder neural network

In this paper a multi-modal unsupervised Deep Learning based algorithm for fault detection is proposed. Such method is applied to real data from a testing procedure implemented on an industrial production line. Both thermal images and current and power measurements coming from industrial refrigerators are collected. The considered dataset is highly unbalanced with the vast majority of samples being healthy. Thermal images are processed via a Deep Convolutional neural network. The features extracted from the thermal images are thus merged to structured data of power, current and temperature. Therefore, a Deep Auto-Encoder is trained on the dataset to signal anomalies corresponding to faults in the refrigerators. Three different methods are trained and compared: (1) an automatic method in which an expert extracts relevant features from thermal images without using the image recognition module; (2) a semi-automatic method where the convolutional neural network is applied to regions of interest within the thermal images selected by an expert operator; (3) a fully automatic method in which the Deep convolutional network processes the whole thermal image without any human intervention. The three methods show comparable results with nevertheless slight differences.