AI-Based Automatic Segmentation of Craniomaxillofacial Anatomy From CBCT Scans for Automatic Detection of Pharyngeal Airway Evaluations in OSA patients

This study aims to generate and also validate an automatic detection algorithm for pharyngeal airway on CBCT data using an AI system which will procure an easy, errorless and fast method. The second aim is to validate the newly developed artificial intelligence system in comparison to commercially available software for 3D CBCT evaluation. A Convolutional Neural Network based machine learning algorithm did the segmentation of the pharyngeal airways in OSA and non-OSA patients. Radiologists used a semi-automatic software to manually determine the airway and their measurements were compared with the AI. OSA patients were classified as minimal, mild, moderate and severe groups and the mean airway volumes were compared. Narrowest points (mm), airway areas (mm2) and airway volumes (cc) of both OSA and non-OSA patients were also compared. There was no statistically significant difference between the manual and Diagnocat measurements in all groups (p>0.05). According to the results of the Diagnocat and manual segmentation, a successful algorithm which can automatically segment the pharyngeal airway from was created which can be used for a swift and precise measurement of pharyngeal airway volume.

'GRAMMARLY' AS A TEACHERS' ALTERNATIVE IN EVALUATING NON -EFL STUDENTS WRITINGS

In evaluating students’ EFL writing, lecturers nowadays can implement corrective evaluation by using an online automatic software. Grammarly is automated online software that is comonly used in EFL writing classes. It is an internet proofreading service that evaluates the correctness grammarl, spelling, punctuation, and vocabulary as well as detects plagiarism. This paper reports research aimed at exploring the use of Grammarly software for evaluating non-EFL students’ writings. This research employed descriptive-qualitative method with students of ITB AAS Indonesia as the data sources. The results of analysis show that in correcting students' language errors, lecturers can evaluate and analyze in details without a lot of correcting efforts or improvements. Grammarly can be considered as a useful tool for lecturers who need to correct non-EFL students’ writings. Grammarly will automatically check or detect the work being typed from various related aspects. Various writing errors made by the students were found in Grammarly’s reports, both in the aspect of correctness and clarity. Correctness is concerned with the mechanical norms in writing, whereas clarity deals with concise and direct language use. The spelling errors found in students’ writings are text inconsistencies, misspelled words, and improper formatting whereas for grammatical there are subject-verb disagreement, passive voice misuse, as well as unclear, wordy, and incomplete sentences. Meanwhile, the errors in punctuation are shown by inappropriate use of punctuation marks in compound/complex sentence, comma misuse within clauses, and improper formatting. It can be concluded that Grammarly can be an alternative for lecturers’ in evaluating non-EFL students' writings.

DESIGNING DISTRIBUTED AUTOMATIC SOFTWARE UPDATE SYSTEM

This article offers the reader a practical guide to the design and development of a distributed client-server system designed to perform automatic software updates using the example of non-trivial functionality. For comparison, examples of existing systems and solutions used to perform software updates are given and justifications for their functional unsuitability in the given circumstances are provided.

Correlation between automatic and manual methods of right ventricle and left atrium strain quantification

Introduction Myocardial strain quantification by speckle-tracking of the right ventricle (RV) and left atrium (LA) can be performed by manual or automatic methods. Purpose The objective of our study is to evaluate the degree of correlation between manual measurement method and automatic software used in our imaging laboratory, in a population of healthy individuals and patients with cardiac transthyretin amyloidosis (ATTR). Methods Fifty-seven individuals were included, 30 patients with ATTR and 27 healthy volunteers, who underwent a transthoracic echocardiogram (TTE) from January to December 2019. Classic echocardiographic parameters and myocardial deformation were obtained according to the ASE/ EACVI guidelines. Global and free wall longitudinal strain of the RV (RVGLS, RVFWLS) and LA global strain (LAGS) analysis were obtained using speckle-tracking with two different software: QLAB Philips 10.7 and AutoSTRAIN Tomtec. Measurements analysis was performed by two experienced echocardiographers. Correlation and reproducibility analysis was performed using Pearson correlation coefficient (PCC) and intraclass correlation coefficient (ICC), respectively. Results Seventy-two percent were male and average age was 63±20 years. Linear correlation of RVGLS, RVFWLS and LAGS measurements with both methods reached statistical significance (table). This correlation was stronger and more reliable in the case of the LAGS. The attached figure shows the correlation between the different software in both groups. Conclusions AutoSTRAIN Tomtec automatic measurement method had higher reliability and correlation comparing to manual measurements performed by QLAB 10.7, especially in LA measurements. The obtained results, the application speed and the less operator's dependence of this automatic software support its routine use for RV and LA strain quantification. FUNDunding Acknowledgement Type of funding sources: None. Correlation and reproducibility results Linear regression lines

Quantitative Breast Density in Contrast-Enhanced Mammography

Contrast-enhanced mammography (CEM) demonstrates a potential role in personalized screening models, in particular for women at increased risk and women with dense breasts. In this study, volumetric breast density (VBD) measured in CEM images was compared with VBD obtained from digital mammography (DM) or tomosynthesis (DBT) images. A total of 150 women who underwent CEM between March 2019 and December 2020, having at least a DM/DBT study performed before/after CEM, were included. Low-energy CEM (LE-CEM) and DM/DBT images were processed with automatic software to obtain the VBD. VBDs from the paired datasets were compared by Wilcoxon tests. A multivariate regression model was applied to analyze the relationship between VBD differences and multiple independent variables certainly or potentially affecting VBD. Median VBD was comparable for LE-CEM and DM/DBT (12.73% vs. 12.39%), not evidencing any statistically significant difference (p = 0.5855). VBD differences between LE-CEM and DM were associated with significant differences of glandular volume, breast thickness, compression force and pressure, contact area, and nipple-to-posterior-edge distance, i.e., variables reflecting differences in breast positioning (coefficient of determination 0.6023; multiple correlation coefficient 0.7761). Volumetric breast density was obtained from low-energy contrast-enhanced spectral mammography and was not significantly different from volumetric breast density measured from standard mammograms.

Epicardial adipose tissue and severe Coronavirus Disease 19

Background Both visceral adipose tissue and epicardial adipose tissue (EAT) have pro-inflammatory properties. The former is associated with Coronavirus Disease 19 (COVID-19) severity. We aimed to investigate whether an association also exists for EAT. Material and methods We retrospectively measured EAT volume using computed tomography (CT) scans (semi-automatic software) of inpatients with COVID-19 and analyzed the correlation between EAT volume and anthropometric characteristics and comorbidities. We then analyzed the clinicobiological and radiological parameters associated with severe COVID-19 (O2 $$\ge$$ ≥ 6 l/min), intensive care unit (ICU) admission or death, and 25% or more CT lung involvement, which are three key indicators of COVID-19 severity. Results We included 100 consecutive patients; 63% were men, mean age was 61.8 ± 16.2 years, 47% were obese, 54% had hypertension, 42% diabetes, and 17.2% a cardiovascular event history. Severe COVID-19 (n = 35, 35%) was associated with EAT volume (132 ± 62 vs 104 ± 40 cm3, p = 0.02), age, ferritinemia, and 25% or more CT lung involvement. ICU admission or death (n = 14, 14%) was associated with EAT volume (153 ± 67 vs 108 ± 45 cm3, p = 0.015), hypertension and 25% or more CT lung involvement. The association between EAT volume and severe COVID-19 remained after adjustment for sex, BMI, ferritinemia and lung involvement, but not after adjustment for age. Instead, the association between EAT volume and ICU admission or death remained after adjustment for all five of these parameters. Conclusions Our results suggest that measuring EAT volume on chest CT scans at hospital admission in patients diagnosed with COVID-19 might help to assess the risk of disease aggravation.