Learning label diffusion maps for semi-automatic segmentation of lung CT images with COVID-19

[Display omitted] • A fully end-to-end deep learning approach for COVID-19 CT image segmentation. • The trained model induces the diffusion of seeds by taking as input a marked slice. • The method learns diffusion maps by predicting edge weights via deep contour learning, • The use of deep contour learning and seeded segmentation as an integrated method. Deep Learning (DL) has become one of the key approaches for dealing with many challenges in medical imaging, which includes lung segmentation in Computed Tomography (CT). The use of seeded segmentation methods is another effective approach to get accurate partitions from complex CT images, as they give users autonomy, flexibility and easy usability when selecting specific targets for measurement purposes or pharmaceutical interventions. In this paper, we combine the accuracy of deep contour leaning with the versatility of seeded segmentation to yield a semi-automatic framework for segmenting lung CT images from patients affected by COVID-19. More specifically, we design a DL-driven approach that learns label diffusion maps from a contour detection network integrated with a label propagation model, used to diffuse the seeds over the CT images. Moreover, the trained model induces the diffusion of the seeds by only taking as input a marked CT-scan, segmenting hundreds of CT slices in an unsupervised and recursive way. Another important trait of our framework is that it is capable of segmenting lung structures even in the lack of well-defined boundaries and regardless of the level of COVID-19 infection. The accuracy and effectiveness of our learned diffusion model are attested to by both qualitative as well as quantitative comparisons involving several user-steered segmentations methods and eight CT data sets containing different types of lesions caused by COVID-19. [ FROM AUTHOR]

Learning Label Diffusion Maps for Semi-Automatic Segmentation of Lung CT Images with COVID-19

Deep Learning (DL) has become one of the key approaches for dealing with many challenges in medical imaging, which includes lung segmentation in Computed Tomography (CT). The use of seeded segmentation methods is another effective approach to get accurate partitions from complex CT images, as they give users autonomy, flexibility and easy usability when selecting specific targets for measurement purposes or pharmaceutical interventions. In this paper, we combine the accuracy of deep contour leaning with the versatility of seeded segmentation to yield a semi-automatic framework for segmenting lung CT images from patients affected by COVID-19. More specifically, we design a DL-driven approach that learns label diffusion maps from a contour detection network integrated with a label propagation model, used to diffuse the seeds over the CT images. Moreover, the trained model induces the diffusion of the seeds by only taking as input a marked CT-scan, segmenting hundreds of CT slices in an unsupervised and recursive way. Another important trait of our framework is that it is capable of segmenting lung structures even in the lack of well-defined boundaries and regardless of the level of COVID-19 infection. The accuracy and effectiveness of our learned diffusion model are attested by both qualitative as well as quantitative comparisons involving several user-steered segmentations methods and eight CT data sets containing different types of lesions caused by COVID-19.

Fuzzy Artificial Intelligence-Based Model Proposal to Forecast Student Performance and Retention Risk in Engineering Education: An Alternative for Handling with Small Data

Understanding the key factors that play an important role in students' performance can assist improvements in the teaching-learning process. As an alternative, artificial intelligence (AI) methods have enormous potential, facilitating a new trend in education. Despite the advances, there is an open debate on the most suitable model for machine learning applied to forecast student performance patterns. This paper addresses this gap, where a comparative analysis between AI methods was performed. As a research hypothesis, a fuzzy inference system (FIS) should provide the best accuracy in this forecast task, due to its ability to deal with uncertainties. To do so, this paper introduces a model proposal based on AI using a FIS. An online survey was carried to collect data. Filling out a self-report, respondents declare how often they use some learning strategies. In addition, we also used historical records of students' grades and retention from the last 5 years before the COVID pandemic. Firstly, two experimental groups were composed of students with failing and passing grades, compared by the Mann-Whitney test. Secondly, an association between the 'frequency of using learning strategies' and 'occurrence of failing grades' was quantified using a logistic regression model. Then, a discriminant analysis was performed to build an Index of Student Performance Expectation (SPE). Considering the learning strategies with greater discriminating power, the fuzzy AI-based model was built using the database of historical records. The learning strategies with the most significant effect on students' performance were lesson review (34.6%), bibliography reading (25.6%), class attendance (23.5%), and emotion control (16.3%). The fuzzy AI-based model proposal outperformed other AI methods, achieving 94.0% accuracy during training and a generalization capacity of 91.9% over the testing dataset. As a practical implication, the SPE index can be applied as a tool to support students' planning in relation to the use of learning strategies. In turn, the AI model based on fuzzy can assist professors in identifying students at higher risk of retention, enabling preventive interventions.