Intraobserver Variability Research Articles

Reliability of left ventricular hemodynamic forces derived from feature-tracking cardiac magnetic resonance.

Hemodynamic forces (HDF) analysis has been proposed as a method to quantify intraventricular pressure gradients, however data on its reliability are still scanty. Thus, the aim of this study is to assess the reliability of HDF parameters derived from cardiac magnetic resonance (CMR). CMR studies of 25 athletes were analysed by two independent observers and then re-analysed by the same observer one week apart. Intraclass Correlation Coefficient (ICC [95% CI]) and Bland-Altman plots were used to assess association, agreement, and bias of the longitudinal (A-B) HDF, transverse (L-S) HDF, and Impulse Angle. The sample size required to detect a relative change in the HDF parameters was also calculated. In terms of inter-observer variability, there was a good correlation for the A-B and L-S (ICC 0.85 [0.67-0.93] and 0.86 [0.69-0.94]; p<0.001 for both, respectively) and a moderate correlation for the Impulse Angle (ICC 0.73 [0.39-0.87]; p = 0.001). For intra-observer variability, A-B and L-S showed excellent correlation (ICC 0.91 [0.78-0.93] and 0.93 [0.83-0.97]; p<0.001 for both, respectively). Impulse Angle presented good correlation (ICC 0.80 [0.56-0.90]; p<0.001). Frame selection and aortic valve area measurements were the most vulnerable step in terms of reliability of the method. Sample size calculation to detect relative changes ranged from n = 1 to detect a 15% relative change in Impulse Angle to n = 171 for the detection of 10% relative change in A-B HDF. The results of this study showed a low inter- and intra-observer variability of HDF parameters derived from feature-tracking CMR. This provides the fundamental basis for their use both in research and clinical practice, which could eventually lead to the detection of significant changes at follow-up studies.

Portuguese version of the Literacy Independent Cognitive Assessment (LICA) instrument in the evaluation of individuals aged 50 years or older with Itabaianinha syndrome.

Individuals with congenital isolated growth hormone deficiency (IGHD) in Northeastern Brazil have a normal lifespan with a prolonged healthspan. We hypothesize that their increased healthspan is accompanied by a reduced cognitive decline during aging. We have recently shown that these individuals have a similar total cognitive function and better attention and executive function than controls. These data were obtained using a Portuguese version of the Literacy Independent Cognitive Assessment (LICA) instrument, whose translation to facilitate cognitive research in Portuguese-speaking countries is described here. In the first stage, a psychologist and a psychiatrist translated the LICA instrument from English into Portuguese, and an English teacher proofread the translation. The second stage included its synthesis and cultural adaptation, carried out by Brazilian authors, and changes in some words and images. The third stage involved an evaluation round with two referees (independent psychologists). The fourth stage involved a back translation of the instrument, which demonstrated > 95% agreement with the original version. The fifth stage included a study to verify the understanding of the questionnaire by responders. In the sixth stage, an endocrinologist and a psychiatrist approved the final Portuguese version of the instrument, which was then administered to 15 individuals with IGHD and 15 controls older than 50 years. The LICA instrument was applied 59 times (5 times in the pilot study, 24 in the variability studies, and 30 in the experimental step). The interobserver and intraobserver variabilities were 99% and 96%, respectively. Cronbach's alpha was 0.76, indicating good reliability. The mean (± standard deviation) duration of the application was 39 ± 8.6 and 48.5 ± 5.8 minutes in literate and illiterate individuals, respectively. The Portuguese version of the LICA instrument was valuable for the cognitive assessment of individuals with Itabaianinha syndrome.

Challenges in the assessment of epithelial dysplasia in oral lichen planus and oral lichenoid lesion: Inter and intra-observer variability of the WHO criteria and binary system.

Assessment of oral epithelial dysplasia is the gold standard for investigating the risk of malignant progression. The World Health Organization (WHO) methods and the binary system have limitations. This study assess the inter- and intra-observer variability of the architectural and cytological criteria and the classification of the presence and degree of epithelial dysplasia in oral lichen planus (OLP) and oral lichenoid lesion (OLL), using both the 2017 WHO criteria and the binary system. The sample consisted of 65 biopsies from lesions classified as OLP and OLL according to the American Academy of Oral and Maxillofacial Pathology (AAOMP) criteria. The histological slides were reevaluated by two oral pathologists. The individual alterations with most inter-observer disagreement were atypical mitotic figures (43.1%), loss of cohesion between epithelial cells (38.5%) and drop shape rete ridges ridges (38.5%). Inter-observer agreement analysis did not show statistically significant agreement regarding the classification of epithelial dysplasia grade by WHO criteria, only regarding the binary system classification (k=0.257; p=0.035). Intra-observer agreement analysis by evaluator 1 showed that the classification of epithelial dysplasia grade according to both methods had statistically significant agreement (k=0.546; p=0.004, k=0.861; p<0.001). Considering evaluator 2, only the evaluation of the WHO system classification showed statistically significant agreement (k=0.593; p=0.010). The evaluation of epithelial dysplasia is subjective and focal changes and inflammatory infiltrate, characteristic of OLP and OLL, can increase the degree of disagreement among evaluators. The binary system presents better inter-observer agreement, while the WHO system presents better intra-observer agreement. Key words:Oral lichen planus, oral lichenoid lesion, oral lichenoid disease, dysplasia, inter-observer variation.

Inter- and intra-observer variability of radial-endobronchial ultrasound image interpretation for peripheral pulmonary lesions.

Radial probe endobronchial ultrasound (R-EBUS) is often utilized in guided bronchoscopy for the diagnosis of peripheral pulmonary lesions. R-EBUS probe positioning has been shown to correlate with diagnostic yield, but overall diagnostic yield with this technology has been inconsistent across the published literature. Currently there is no standardization for R-EBUS image interpretation, which may result in variability in grading concentricity of lesions and subsequently procedure performance. This was a survey-based study evaluating variability among practicing pulmonologists in R-EBUS image interpretation. R-EBUS images from peripheral bronchoscopy cases were sent to 10 practicing Interventional Pulmonologists at two different time points (baseline and 3 months). Participants were asked to grade the images as concentric, eccentric, or no image. Cohen's Kappa-coefficient was calculated for inter- and intra-observer variability. A total of 100 R-EBUS images were included in the survey. There was 100% participation with complete survey responses from all 10 participants. Overall kappa-statistic for inter-observer variability for Survey 1 and 2 was 0.496 and 0.477 respectively. Overall kappa-statistic for intra-observer variability between the two surveys was 0.803. There is significant variability between pulmonologists when characterizing R-EBUS images. However, there is strong intra-rater agreement from each participant between surveys. A standardized approach and grading system for radial EBUS patterns may improve inter-observer variability in order to optimize our clinical use and research efforts in the field.

Intra and Inter-Observer Variability in the Measurement of Anterior Chamber Depth Diameter Using B-Mode Ultrasonography

Introduction: B-mode ultrasonography, a non-invasive imaging modality, plays a pivotal role in evaluating ocular structures such as anterior chamber depth. Its measurement is crucial for ocular health assessment and managing conditions like glaucoma cataract and refractive surgery planning. However, ultrasonography's operator-dependent nature impacts diagnostic sensitivity and specificity.&#x0D; Aims: This study investigates intra-observer and inter-observer variability in ultrasonographic ACD measurements and its implications.&#x0D; Methods: A prospective observational study was conducted at Nepalgunj Medical College and Teaching Hospital in September 2023. Healthy volunteers over 17 years of age were enrolled, with ethical approval and informed consent. Ultraso nography was performed using a 7.5 MHz linear transducer, involving a radiologist with over 5 years of experience and a radiology trainee. Reproducibility was assessed using intraclass correlation coefficients (ICC).&#x0D; Results: The study included 150 participants (83 males, 67 females) with a mean age of 25.7 years. ACD demonstrated a linear correlation with age. The mean right ACD was 2.824 mm (range: 2.175 mm - 4.175 mm) and the mean left ACD was 2.826 mm (range: 2.20 mm - 4.175 mm). Age had a significant linear correlation with ACD (p &lt;0.001). Intra-observer variability for both observers showed statistically significant linear correlations (p &lt;0.001), with high Pearson correlation coefficients. Inter-observer variability in mean ACD also exhibited a significant correlation (p&lt;0.01) with excellent agreement.&#x0D; Conclusion: This study highlights the importance of assessing intra and inter-observer variability in ACD measurements by ultrasonography. The findings provide valuable insights for ocular health management, particularly in patients with elevated intraocular pressure. Standardizing measurement techniques can improve diagnostic accuracy and patient care.

Psychometric Performance of the Stony Brook Scar Evaluation Scale and SCAR-Q Questionnaire in Dutch Children after Pediatric Surgery.

Introduction: The growing population of survivors following pediatric surgery emphasizes the importance of long-term follow-up. The impact of surgical scars on daily life can be evaluated through patient-reported outcome measurements. The Stony Brook Scar Evaluation Scale (SBSES) and SCAR-Q questionnaire are two interesting instruments for this purpose. We evaluated their psychometric performance in Dutch children after pediatric surgery. Methods: After English-Dutch translation, we evaluated-following the COSMIN guidelines-the feasibility, reliability (internal and external), and validity (construct, criterion, and convergent) of the SBSES and SCAR-Q in Dutch patients < 18 years old with surgical scars. Results: Three independent observers completed the SB for 100 children (58% boys, median age 7.3 (IQR 2.5-12.1) years) in whom surgery had been performed a median of 2.8 (0.5-7.9) years ago. Forty-six of these children (61% boys, median age 12.1 (9.3-16.2) years) completed the SCAR-Q. Feasibility and internal reliability (Cronbach's alpha > 0.7) was good for both instruments. For the SB, external reliability was poor to moderate (interobserver variability: ICC 0.46-0.56; intraobserver variability: ICC 0.74). For the SCAR-Q, external reliability was good (test-retest agreement: ICC 0.79-0.93). Validity tests (construct, criterion, and convergent) showed poor to moderate results for both instruments. Conclusions: The Dutch-translated SBSES and SCAR-Q showed good feasibility and internal reliability. External reliability and validity were likely affected by differences in conceptual content between the questionnaires. Combining them would provide insight in the impact of scars on patients. Implementation of these instruments in longitudinal follow-up programs could provide new insights into the long-term psychological outcome after pediatric surgery.

Crowdsourcing image segmentation for deep learning: integrated platform for citizen science, paid microtask, and gamification.

Segmentation is crucial in medical imaging. Deep learning based on convolutional neural networks showed promising results. However, the absence of large-scale datasets and a high degree of inter- and intra-observer variations pose a bottleneck. Crowdsourcing might be an alternative, as many non-experts provide references. We aim to compare different types of crowdsourcing for medical image segmentation. We develop a crowdsourcing platform that integrates citizen science (incentive: participating in the research), paid microtask (incentive: financial reward), and gamification (incentive: entertainment). For evaluation, we choose the use case of sclera segmentation in fundus images as a proof-of-concept and analyze the accuracy of crowdsourced masks and the generalization of learning models trained with crowdsourced masks. The developed platform is suited for the different types of crowdsourcing and offers an easy and intuitive way to implement crowdsourcing studies. Regarding the proof-of-concept study, citizen science, paid microtask, and gamification yield a median F-score of 82.2, 69.4, and 69.3 % compared to expert-labeled ground truth, respectively. Generating consensus masks improves the gamification masks (78.3 %). Despite the small training data (50 images), deep learning reaches median F-scores of 80.0, 73.5, and 76.5 % for citizen science, paid microtask, and gamification, respectively, indicating sufficient generalizability. As the platform has proven useful, we aim to make it available as open-source software for other researchers.

Left ventricular fibrosis in arrhythmic mitral valve prolapse: quantification and comparison of semi-automated techniques assessed by cardiac magnetic resonance

PurposeLeft ventricular (LV) fibrosis has a key role in arrhythmogenesis in patients with mitral valve prolapse (MVP). Cardiac magnetic resonance identifies LV fibrosis by using late gadolinium enhancement (LGE) technique. LGE assessment and quantification in patients with MVP lacks of standardization protocols.Methods66 MVP patients with normal systolic function and without significant regurgitation were enrolled. Semi-automated gray-scale thresholding techniques using full width at half maximum (FWHM) and 2, 3 and 5 standard deviation (SD) above the remote myocardium were used and compared with the visual assessment, considered as the gold standard.ResultsLGE was identified in 41 MVP patients (62%) and quantified. The mean quantity of LGE visually assessed was 2.40 ± 1.07% or 1.40 ± 0.82 g. With FWHM, LGE resulted 3.56 ± 1.23% or 1.99 ± 1.13 g. Using thresholding, the mean LGE quantity was 9.2 ± 3.1% or 4.82 ± 2.28 g for 2-SD, 5.72 ± 1.75% or 3.06 ± 1.47 g for 3-SD and 2.36 ± 0.99% or 1.29 ± 0.79 g for 5-SD. The 5-SD measurement in percentage demonstrated a good correlation with LGE quantification visually assessed (2.40 ± 1.07 vs. 2.363 ± 0.9909, p = 0.543). When compared with the gold standard, the 5-SD threshold quantification, both in percentage and in grams, revealed the least intra-observer (respectively, ICC: 0.976 and 0.966) and inter-observer variability (respectively ICC: 0.948 and 0.935).ConclusionThe 5-SD gray-scale threshold technique in percentage revealed the best correlation with the visual assessment and an optimal reproducibility in MVP patient.

The Influence of Late Gadolinium Enhancement Cardiac Magnetic Resonance Image Analysis Imprecision on Myocardial Damage Quantification in Patients with Myocarditis: A Pilot Study

Background: Myocardial damage in myocarditis is assessed through late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR). Variability in quantifying myocarditis extent results from imprecise image segmentation and inconclusive data on quantification method selection. To improve analysis precision, segmentation steps are systematically ranked based on their inherent risks of error. Additionally, data on two distinct quantification methods are presented. Methods: Using newly developed software, four experts analyzed five LGE-CMR left ventricular (LV) short-axis (SAx) images of myocarditis patients in three sessions. Regions of interest (ROIs) (myocardial (ROImyoc), reference (ROIref), and exclusion region (ROIexcl)) were identified and used to calculate LGE extent with 3σ (intensity above three standard deviations (σ) in reference) and the full width at half maximum (FWHM) method (intensity above 50% of maximum signal in reference). The reference LGE extent was calculated and the influence of the ROIs on LGE extent variability was determined. Interobserver and intraobserver variability were evaluated as 1-intraclass correlation coefficient (ICC). Results: LGE extent variability was 6.2 ± 0.6% for 3σ and 4.0 ± 0.6% for FWHM. The contributions of ROImyoc, ROIref, and ROIexcl were 1.5 ± 0.2%, 2.7 ± 0.4%, and 2 ± 0.3%, respectively, for 3σ, and 1.1 ± 0.1%, 1.6 ± 0.4%, and 1.3 ± 0.3%, respectively, for FWHM. LGE extent was lower in FWHM. Interobserver variability was 0.56 for 3σ and 0.43 for FWHM. The intraobserver variability was higher for the 3σ method in all four observers. Conclusion: ROIref selection contributed most to LGE extent variability. FWHM yielded lower LGE extent and lower inter- and intraobserver variability. Due to low statistical significance, the findings are only partially confirmed.

The role of Chest HRCT in diagnosis active tuberculosis &amp; lung destruction

Tuberculosis is a public health problem caused by Mycobacterium tuberculosis. In 2021 there will be 10.6 million cases in the world, and Indonesia ranks 2nd with 700,000 cases. In 2022, there will be 17,303 cases in North Sumatra Province and 2,430 cases in Medan City. In current practice, evaluation and diagnosis of active tuberculosis relies on bacteriological examination and Chest Radiographs. However, Chest Radiographs have limited specificity and high intraobserver and interobserver variability. HRCT is also not widely used as a routine option for patients suspected of suffering from tuberculosis, even though HRCT has high accuracy in detecting tuberculosis. We present a case of a patient with active pulmonary tuberculosis and lung destruction e.c. advanced pulmonary tuberculosis (MDR-Tuberculosis) (declared cured in 2018 after 2 years of treatment). This patient was evaluated with chest radiograph and chest HRCT. In this case, the initial findings on the Chest Radiograph, showed the impression of inactive tuberculosis but on further examination with chest HRCT, there was a tree in bud image which indicated active tuberculosis. This shows that there are limitations in diagnosing tuberculosis activity using Chest Radiographs. Due to the significant role of HRCT in the diagnosis of tuberculosis activity, it is necessary to consider the use of HRCT in the evaluation of patients with tuberculosis.

Determination of Values of the Tricuspid Annular Plane Systolic Excursion (TAPSE) In Bangladeshi Adult Patient With or Without LV Dysfunction

Background: Tricuspid annular plane systolic excursion (TAPSE) is an Echocardiographic measure that allows us to assess right ventricular systolic function and it correlates well with reference techniques like Cardiac Magnetic Resonance Imaging. TAPSE measurement is common in adults. Objective: Our objective was to determine values of TAPSE in Bangladeshi adults with or without LV systolic dysfunction and to determine the relationship of these values with age, sex, RVIDd, LVEF &amp; EPSS. Method: This was an prospective observational study in patients undergoing comprehensive transthoracic echocardiography for any indication. From April 2022 to April 2023 we prospectively enrolled 100 adult (from 30 to age 80 years) of both sexes who presented to the Cardiology Clinic of LABAID Cardiac Hospital. We performed a complete transthoracic echocardiography study. We measured TAPSE in 2-dimensional M-mode echocardiograms from the apical 4- chamber view, positioning the cursor on the lateral tricuspid annulus near the free RV wall and aligning it as close as possible to the apex of the heart . The mean values were taken by at least 2 measurements for reducing interobserver and intraobserver variability’s. Patients with confirmed congenital &amp; valvular heart disease were excluded. Results: Mean TAPSE values were 19.15±3.87cm irrespective of LVEF, with no significant differences between sexes, 18.45±3.801 in male , 19.94±3.853 in female (P=.056). TAPSE value was 22.00±1.581 in person with normal LVEF &amp; 16.77±2.455 in person with reduced LVEF ( P&lt;.001) A statistically significant positive correlation of LVEF (r=.813) and significant negative correlation of EPSS (r=-.639) were observed with TAPSE (p&lt;0.001). But no significant correlation of TAPSE was found between age (r=-.185)). ), gender (r=.192) &amp; RVIDd (r= -.063) (p &gt;0.05). Multivariate analysis confirmed these correlations and the interactions between variables (LVEF &amp; EPSS). Graphs of estimated population-based TAPSE values adjusted by age and LV function are provided. Conclusion: We determined values of TAPSE in Bangladeshi adult population with or without LV systolic dysfunction and assessed relationship of these values with age, sex, RVIDd, LVEF &amp; EPSS. The TAPSE measurement was reproducible and associated directly with LV systolic function. These reference values could guide decision making in daily clinical practice. Bangladesh Heart Journal 2023; 38(2): 92-101

Development of birth weight estimation model for Ethiopian population from sonographic evaluation

BackgroundFetal birth weight (FBW) estimation involves predicting the weight of a fetus prior to delivery. This prediction serves as a crucial input for ensuring effective, accurate, and appropriate obstetric planning, management, and decision-making. Typically, there are two methods used to estimate FBW: the clinical method (which involves measuring fundal height and performing abdominal palpation) or sonographic evaluation. The accuracy of clinical method estimation relies heavily on the experience of the clinician. Sonographic evaluation involves utilizing various mathematical models to estimate FBW, primarily relying on fetal biometry. However, these models often demonstrate estimation errors that exceed acceptable levels, which can result in inadequate labor and delivery management planning. One source of this estimation error is sociodemographic variations between population groups in different countries. Additionally, inter- and intra-observer variability during fetal biometry measurement also contributes to errors in FBW estimation.MethodsIn this research, a novel mathematical model was proposed through multiple regression analysis to predict FBW with an accepted level of estimation error. To develop the model, population data consisting of fetal biometry, fetal ultrasound images, obstetric variables, and maternal sociodemographic factors (age, marital status, ethnicity, educational status, occupational status, income, etc.) of the mother were collected. Two approaches were used to develop the mathematical model. The first method was based on fetal biometry data measured by a physician and the second used fetal biometry data measured using an image processing algorithm. The image processing algorithm comprises preprocessing, segmentation, feature extraction, and fetal biometry measurement.ResultsThe model developed using the two approaches were tested to assess their performance in estimating FBW, and they achieved mean percentage errors of 7.53% and 5.89%, respectively. Based on these results, the second model was chosen as the final model.ConclusionThe findings indicate that the developed model can estimate FBW with an acceptable level of error for the Ethiopian population. Furthermore, this model outperforms existing models for FBW estimation. The proposed approach has the potential to reduce infant and maternal mortality rates by providing accurate fetal birth weight estimates for informed obstetric planning.

Enhancing Precision and Efficiency in Knee Arthroplasty: A Comparative Analysis of Computer-Assisted Measurements with a Novel Software Tool versus Manual Measurements for Lower Leg Geometry.

(1) Background: The aim of this prospective study was to evaluate measurement software in comparison with manual measurements using inter-observer and intra-observer variability on radiographs in the preoperative planning of total knee arthroplasty. (2) Methods: Two independent observers retrospectively measured the mechanical lateral proximal femoral angle (mLPFA), the mechanical lateral distal femoral angle (mLDFA), the joint line convergence angle (JLCA), the mechanical medial proximal tibial angle (mMPTA), the mechanical lateral distal tibial angle (mLDTA), the hip-knee angle or mechanical tibial-femoral axis angle (HKA), and the anatomical-mechanical angle (AMA) on 55 long-leg anteroposterior radiographs manually twice, followed by measurements using dedicated software. Variability between manual and computer-aided planning was assessed, and all measurements were performed a second time after 14 days in order to assess intra-observer variability. (3) Results: Concerning intra-observer variability, no statistically significant difference was observed regarding the software-based measurements. However, significant differences were noted concerning intra-observer variability when measuring the mLDFA and AMA manually. Testing for statistical significance regarding variability between manual and software-based measurements showed that the values varied strongly between manual and computer-aided measurements. Statistically significant differences were detected for mLPFA, mLDFA, mMPTA, and mLPTA on day 1, and mLPFA, mMPTA, and mLPTA on day 15, respectively. (4) Conclusions: Preoperative planning of leg axis angles and alignment using planning software showed less inter- and intra-observer variability in contrast to manual measurements, and results differed with respect to manual planning. We believe that the planning software is more reliable and faster, and we would recommend its use in clinical settings.

The use of virtual reality to assess the bony landmarks at the knee joint – The role of imaging modality and the assessor's experience

BackgroundAt present, extended reality technologies such as virtual reality (VR) have gained popularity in orthopedic surgery. The first aim of this study was to assess the precision of VR and other imaging modalities – computed tomography (CT), magnetic resonance imaging (MRI) – to localize bony landmarks near the knee joint. Secondly, the impact of the educational level of the assessor – medical master students, orthopedic residents, and orthopedic surgeons – on the precision with which landmarks near the knee joint could be localized was analyzed. MethodsWe included a total of 77 participants: 62 medical master students, 10 orthopedic residents, and 5 orthopedic surgeons to analyze three cadaver legs. Every participant localized a series of sixteen bony landmarks on six different imaging modalities (CT, MRI, 3D-CT, 3D-MRI, VR-CT, VR-MRI). ResultsConcerning the imaging modality, the inter- and intra-observer variability were lowest for 3D and VR, higher for MRI (respectively 7.6 mm and 6.9 mm), and highest for CT (respectively 9 mm and 8.7 m).Concerning the educational level of the assessor, inter- and intra-observer variability in VR were lowest for surgeons, (respectively 3.2 mm and 3.6 mm), higher for residents (respectively 5.9 mm and 6.5 mm) and medical students (respectively 5.9 mm and 5.8 mm). ConclusionsVR can be considered a reliable imaging technique. Localization of landmarks tends to be more precise in VR and on 3D than on conventional CT and MRI images. Furthermore, orthopedic surgeons localize landmarks more precisely than orthopedic residents and medical students in VR.

Design and performance analysis of Cobb angle measurement from X-ray images

Scoliosis seems to be the most frequently used type of spinal abnormality. Cobb angle measurement essentially relates to the quantification of the spinal stenosis in degrees, also it is a globally standardized approach for assessing scoliosis. Based on a recent survey, there really is no accurate and comprehensive technique for calculating the Cobb angle programmatically. This problem is crucial in the medical field pertaining to Cobb angle measurement to identify the exact position of Cobb in X-ray images. However, multiple investigations have demonstrated that there is inter and intra-observer variance when assessing the Cobb angle physically. The goal is to create a computer-assisted solution to reduce user-based Cobb angle measuring errors. The preprocessing filters and semi-automatic methods determine the overall architectural curved spine. Using the Cobb technique, results are inaccuracies and improper in the estimation of a scoliotic curvature's peak or bottom vertebra. The curve-fitting approach was used in this investigation to reduce uncertainty. Every patient had a digitally recorded poster anterior radiography image. The polynomial estimation is fitted using user-defined curvature midpoints that correspond to vertebral intersection points. The Cobb angle is computed by taking the first characteristic of the fitted polynomial function and dividing it by the number of vertebrae.

Developing Topics.

Alzheimer's disease (AD) is highly related to brain amyloid and tau deposition. PET images evaluate these patterns of amyloid and tau proteinopathy noninvasively. Tau PET is less familiar than amyloid PET to clinicians and researchers, it may occur inter- and intra-observer variability to determine abnormalities in visual rating. We purposed to find optimal thresholds of tau PET, compared with visual assessment, and suggest the most appropriate standard of tau PET assessment using the assist of automated quantitative analysis of PET. 101 participants were enrolled in tau (18F-flortaucipir), amyloid (18F-flutemetamol) PET, and a three-dimension T1-weighted MR image study. Each regional SUVR was calculated using SCALE PET version 1.2 (Neurophet Inc., Seoul, Republic of Korea.). Each regions-of-interest (ROI)-mean SUVR of reference region (whole cerebellum (CW) and cerebellar gray matter (CG)) and quantitative Tau PET positivity was compared with visual rating results of a nuclear medicine physician. We compared accuracy of tau PET positivity driven from 1) total Braak ROI, 2) a priori ROI, and 3) whole cerebral cortical mean SUVR above 1.65 (CW, CG) or 1.30 (CG). The highest accuracy was observed 0.832 with above 1.65 mean SUVR in a priori ROI using CG reference region. False negative (FN), where participants classified as negative with mean SUVR when visually positive (FN), was 26.7%. These FN participants were mostly identified as borderline patients who were categorized to Braak III/IV (3) or V/VI (24) in visual rating. All the participants classified as tau positive in the quantitative method were amyloid positive while 54.1% were in the tau negative group. Also, the patient group composed of AD and mild cognitive impairment (MCI) participants was 96.3% in positive, behalf 55.4% in negative group. Both visual rating and quantitative classification showed a significant difference between negative and positive groups (p <0.001). Visual assessment may provide more sensitive outcome than quantitative estimation, i.e., mean SUVR. Our quantitative methods of tau PET showed high accuracy with visual assessment in the regions of early Braak stage, i.e., Braak I/II/III/IV. In the future, there needed to develop more sensitive cutoff values or quantitative methods for tauopathy deposition observation.

Real-World Demonstration of AI to Clinical Cytometry for Rapid and Reliable Decision Support and Automated Reporting of Blood Cancers Including B-NHL and Acute Leukemias

Introduction: Flow cytometry is an integral part of routine diagnostics for hematological malignancies. Unfortunately, a decreasing number of skilled operators have to cope with increasing case volumes. The current methods are also known to produce inter- and intra-observer variability. Aim: To reduce labor time, dependency on expert knowledge and interpretation variability, we aimed to build and validate clinical-grade decision support software in the routine using machine learning (ML) for clinical cytometry. Methods: We obtained ≈100k flow cytometry cases from multiple centers, taken from clinical routine and annotated by expert users using multiple diagnostic methods. These cases span the full spectrum of hematological malignancy, including 47% non-malignant reference samples. As the data were not measured using harmonized protocols, we developed data pooling techniques to merge the data, including synthetic imputation of missing markers. After data processing, we applied multi-layer artificial neural networks to predict sub-type or non-malignancy (B-NHL and acute leukemias, ten classes at time of writing), directly from the data without any human preprocessing. Transfer learning was then used to fine-tune the model to a particular protocol for which the model could be used in routine. Besides sub-typing recommendations, users need to report cell population frequencies, traditionally measured using manual ‘gating’. We trained separate supervised ML models to classify single cells based on a combination of curated and refined routine gating data, expanded with in-house annotations produced by a highly trained cytometrist. To enable clinicians to receive the diagnostic recommendations and produce reports for the treating doctor, we built a CE-IVD web-application (“hema.to”) in which selected users could upload cytometry data, inspect the raw data including abnormalities in the immunophenotype, and audit the ML recommendations against the WHO criteria. The user feedback from each case was recorded, from which the consistency of the ML recommendations with expert opinion could be compared. Before the product was integrated in the clinical routine, a ‘dry run’ retrospective two-arm, four-center blinded clinical study using 96 randomly selected non-overlapping B-NHL cases revealed that analysis time was reduced by &amp;gt;2x and accuracy saw a slight (non-significant) increase against a gold standard as compared to the control arm with a traditional workflow. Results: A worst-case sub-typing performance was computed by testing the model against historic data without expert supervision, and found an f1 score of &amp;gt;90% and sensitivity of 96%. The ML models for acute leukemias specifically show an f1 score of ≈90%, a false positive rate of ≈1.5% and false negative rate of 2-6%. These quality metrics demonstrate that the system can be used for both screening and sub-typing. We therefore integrated hema.to into the routine for B-NHL diagnosis at Result Laboratorium (Dordrecht) and the HpH (Hamburg) since January, for the latter including a deep integration into the laboratory database with automated reporting of expressed CD markers. To date, over a thousand routine cases have been analyzed using the system. We then measured the relationship between predicted sub-typing confidence of the model and agreement with the final expert judgment. The data showed that, on average, confidence≈accuracy ( R2=0.94), indicating that the model is neither over- nor underconfident. As such, the model is a “trustworthy” classifier. Expert agreement over time revealed that the quality of the diagnostic recommendations has been stable over time despite a change of device type and was consistently in excess of 90% across various metrics, including a 100% specificity during the four most recent recorded months. Conclusions: We've shown that machine learning can offer decision support and automated reporting for screening and classifying blood cancers from flow cytometry in a real-world routine setting. Our next focus is to expand the number of indications and increase the model sensitivity to small pathological populations, including measurable residual disease. These results represent major strides towards decision support software for any lab and all hematological neoplasms. Such a software will not only speed up and simplify diagnostic workflows, but also improve the quality of the analysis.

An automatic Darknet-based immunohistochemical scoring system for IL-24 in lung cancer

Immunohistochemical (IHC) detection is of critical importance in the pathological diagnosis of lung cancer. Interleukin-24 (IL-24) is a significant predictive and prognostic marker in IHC detection, which can help to characterize the tumor and predict the clinical course. To determine the score of the IL-24 expression, pathologists will assess the positive area and staining intensity through a microscope and then use a semi-quantitative assessment method to assign a score for each IHC image of lung cancer. However, this process is a time-consuming, imprecise, and subjective process, which can result in inter- and intra-observer discrepancies. Meanwhile, The performance of computer-aided diagnosis (CAD) systems for IHC scoring depends on the quality of manually extracted features. For example, many advanced methods require precise extraction of features of cell nuclei and membranes. However, the complex background and overlapping cells in lung cancer IHC images present significant challenges for segmentation, which can affect the accuracy of the final system's predicted scoring results. In this paper, an automatic Darknet-based IHC scoring system for IL-24 in lung cancer is proposed. Firstly, the original IHC images of lung cancer are blocked and the features are concentrated by a “block attention mechanism, which can reduce the computational burden of the analysis of millions of pixels in IHC images. The blocked images are then inputted into a Darknet-based scoring network, which incorporates a novel feature extraction backbone and loss function to obtain the final scores. To the best of our knowledge, this is the first end-to-end system that directly outputs a clinical score using lung cancer IL-24 IHC images as input. We have constructed a dataset of 5000 manually annotated IL-24 IHC images of lung cancer obtained from the Institute of Life Science and Bioengineering at Beijing Jiaotong University. We will present experimental results to demonstrate the feasibility of our proposed method, which can greatly assist with the clinical diagnosis and treatment of lung cancer.

Orientation of the Scapula in the Standing Position.

A new ultrasound-based device is proposed to non-invasively measure the orientation of the scapula in the standing position to consider this parameter for Total Shoulder Arthroplasty. The aim of this study was to assess the accuracy and reliability of this device. Accuracy was assessed by comparing measurements made with the ultrasound device to those acquired with a three-dimensional (3D) optical localization system (Northern Digital, Canada) on a dedicated mechanical phantom. Three users performed 10 measurements on three healthy volunteers with different body mass (BMI) indices to analyze the reliability of the device by measuring the intra and interobserver variabilities. The mean accuracy of the device was 0.9°± 0.7 (0.01-3.03), 1.3°± 0.8 (0.03-4.55), 1.9°± 1.5 (0.05-5.76), respectively, in the axial, coronal, and sagittal planes. The interobserver and intraobserver variabilities were excellent whatever the BMI and the users experience. The device is accurate and reliable enough for the measurement of the scapula orientation in the standing position.

SpineHRformer: A Transformer-Based Deep Learning Model for Automatic Spine Deformity Assessment with Prospective Validation.

The Cobb angle (CA) serves as the principal method for assessing spinal deformity, but manual measurements of the CA are time-consuming and susceptible to inter- and intra-observer variability. While learning-based methods, such as SpineHRNet+, have demonstrated potential in automating CA measurement, their accuracy can be influenced by the severity of spinal deformity, image quality, relative position of rib and vertebrae, etc. Our aim is to create a reliable learning-based approach that provides consistent and highly accurate measurements of the CA from posteroanterior (PA) X-rays, surpassing the state-of-the-art method. To accomplish this, we introduce SpineHRformer, which identifies anatomical landmarks, including the vertices of endplates from the 7th cervical vertebra (C7) to the 5th lumbar vertebra (L5) and the end vertebrae with different output heads, enabling the calculation of CAs. Within our SpineHRformer, a backbone HRNet first extracts multi-scale features from the input X-ray, while transformer blocks extract local and global features from the HRNet outputs. Subsequently, an output head to generate heatmaps of the endplate landmarks or end vertebra landmarks facilitates the computation of CAs. We used a dataset of 1934 PA X-rays with diverse degrees of spinal deformity and image quality, following an 8:2 ratio to train and test the model. The experimental results indicate that SpineHRformer outperforms SpineHRNet+ in landmark detection (Mean Euclidean Distance: 2.47 pixels vs. 2.74 pixels), CA prediction (Pearson correlation coefficient: 0.86 vs. 0.83), and severity grading (sensitivity: normal-mild; 0.93 vs. 0.74, moderate; 0.74 vs. 0.77, severe; 0.74 vs. 0.7). Our approach demonstrates greater robustness and accuracy compared to SpineHRNet+, offering substantial potential for improving the efficiency and reliability of CA measurements in clinical settings.