New patch-based strategy for COVID-19 automatic identification using chest x-ray images.

Purpose: The development of a robust model for automatic identification of COVID-19 based on chest x-rays has been a widely addressed topic over the last couple of years; however, the scarcity of good quality images sets, and their limited size, have proven to be an important obstacle to obtain reliable models. In fact, models proposed so far have suffered from over-fitting erroneous features instead of learning lung features, a phenomenon known as shortcut learning. In this research, a new image classification methodology is proposed that attempts to mitigate this problem. Methods: To this end, annotation by expert radiologists of a set of images was performed. The lung region was then segmented and a new classification strategy based on a patch partitioning that improves the resolution of the convolution neural network is proposed. In addition, a set of native images, used as an external evaluation set, is released. Results: The best results were obtained for the 6-patch splitting variant with 0.887 accuracy, 0.85 recall and 0.848 F1score on the external validation set. Conclusion: The results show that the proposed new strategy maintains similar values between internal and external validation, which gives our model generalization power, making it available for use in hospital settings. Supplementary Information: The online version contains supplementary material available at 10.1007/s12553-022-00704-4.

Inteligencia artificial: una herramienta en la imagenología para los pacientes positivos a la COVID-19 / Artificial intelligence: an imaging tool for COVID-19 positive patients

RESUMEN Introducción: la enfermedad por SARS-Cov-2 refuerza la importancia del uso de las nuevas tecnologías de la información y las comunicaciones en función del desarrollo e implementación de sistemas de inteligencia artificial que favorecen el diagnóstico. Objetivo: describir la posibilidad del uso de la inteligencia artificial como una herramienta en la imagenología para los pacientes positivos a la COVID-19. Métodos: se realizó una revisión de fuentes bibliográficas en Infomed, SciELO, PubMed y Google Académico, comprendidas en los años 2015 al 2020 con el uso de palabras claves: coronavirus, COVID-19, neumonía, radiografía e inteligencia artificial. Se seleccionaron 28 documentos por su pertinencia en el estudio. Desarrollo: la creación de sistemas de inteligencia artificial que ayuden al diagnóstico médico requiere un enfoque interprofesional de la ciencia y constituye una de las líneas de trabajo en Cuba durante la pandemia. Una condición indispensable para la introducción de la inteligencia artificial en el diagnóstico radiológico es la capacitación que deben recibir los médicos para interactuar con ella, a través de un proceso formativo que incluya una evaluación y explicación de la calidad de los datos asociada tanto al aprendizaje como a las nuevas predicciones. Conclusiones: la utilización de inteligencia artificial mejorará el rendimiento del radiólogo para distinguir la COVID-19; la integración de estas tecnologías en el flujo de trabajo clínico de rutina puede ayudar a los radiólogos a diagnosticar con precisión.

ABSTRACT Introduction: SARS-Cov-2 disease reinforces the importance of the use of new information and communication technologies based on the development and implementation of artificial intelligence systems that favor diagnosis. Objective: to describe the possibility of using artificial intelligence as a tool in imaging for COVID-19 positive patients. Methods: a review of bibliographic sources was carried out in Infomed, SciELO, PubMed and Google Scholar, from 2015 to 2020 with the use of keywords: coronavirus, COVID-19, pneumonia, radiography and artificial intelligence. 28 documents were selected for their relevance in the study. Development: the creation of artificial intelligence systems that help medical diagnosis requires an interprofessional approach to science and constitutes one of the lines of work in Cuba during the pandemic. An essential condition for the introduction of artificial intelligence in radiological diagnosis is the training that doctors must receive to interact with it, through a training process that includes an evaluation and explanation of the quality of the data associated with both learning and to new predictions. Conclusions: the use of artificial intelligence will improve the radiologist's performance to distinguish COVID-19; integrating these technologies into routine clinical workflow can help radiologists diagnose accurately.

Current limitations to identify covid-19 using artificial intelligence with chest x-ray imaging (part ii). The shortcut learning problem.

Since the outbreak of the COVID-19 pandemic, computer vision researchers have been working on automatic identification of this disease using radiological images. The results achieved by automatic classification methods far exceed those of human specialists, with sensitivity as high as 100% being reported. However, prestigious radiology societies have stated that the use of this type of imaging alone is not recommended as a diagnostic method. According to some experts the patterns presented in these images are unspecific and subtle, overlapping with other viral pneumonias. This report seeks to evaluate the analysis the robustness and generalizability of different approaches using artificial intelligence, deep learning and computer vision to identify COVID-19 using chest X-rays images. We also seek to alert researchers and reviewers to the issue of "shortcut learning". Recommendations are presented to identify whether COVID-19 automatic classification models are being affected by shortcut learning. Firstly, papers using explainable artificial intelligence methods are reviewed. The results of applying external validation sets are evaluated to determine the generalizability of these methods. Finally, studies that apply traditional computer vision methods to perform the same task are considered. It is evident that using the whole chest X-Ray image or the bounding box of the lungs, the image regions that contribute most to the classification appear outside of the lung region, something that is not likely possible. In addition, although the investigations that evaluated their models on data sets external to the training set, the effectiveness of these models decreased significantly, it may provide a more realistic representation as how the model will perform in the clinic. The results indicate that, so far, the existing models often involve shortcut learning, which makes their use less appropriate in the clinical setting.

Caracterización clínica y radiológica de pacientes positivos a la COVID-19 / Clinical and radiological characterization of COVID-19 positive patients

Introducción: Desde el surgimiento de los primeros casos en la pandemia de la COVID-19, se ha desarrollado una carrera vertiginosa en crear un espacio de investigación para el diagnóstico, tratamiento y control de la enfermedad. Objetivo: Describir las características clínicas y radiológicas de los pacientes con la COVID-19. Métodos: Se realizó un estudio descriptivo, en el período comprendido de marzo a octubre del año 2020, se estudiaron 404 pacientes de todas las edades, ingresados, con diagnóstico confirmado con PCR en tiempo real. Las variables utilizadas fueron: edad, sexo, síntomas y radiografía del tórax. Resultados: El 54,5 por ciento de los pacientes fueron del sexo femenino y entre ellos asintomáticos el 55,9 por ciento; el 36,9 por ciento tenía entre 40 a 59 años de edad, en los menores de 20 años, el 64,9 por ciento no presentó síntomas de la enfermedad al ingreso. Estuvieron asintomáticos el 53,5 por ciento; el 76,6 por ciento de las radiografías positivas correspondieron a los sintomáticos, la tos fue el síntoma más frecuente. La mayor positividad en la radiografía del tórax se encontró en los pacientes mayores de 60 años, se observó como patrón más frecuente, la opacidad en velo, de distribución periférica. Conclusiones: Predominan los pacientes asintomáticos, la positividad de las radiografías es mayor en los ancianos(AU)

Introduction: Since the emergence of the first cases of COVID-19 pandemic, a dizzying race has developed in creating a research space for the diagnosis, treatment and control of the disease. Objective: To describe the clinical and radiological characteristics of patients with COVID-19. Methods: A descriptive study was carried out, in the period from March to October 2020, 404 patients of all ages, admitted, with confirmed diagnosis with real-time PCR, were studied. The variables used were: age, sex, symptoms and chest X-ray. Results: 54.5 percent of the patients were female and 55,9 percent of them were asymptomatic, 36,9 percent were between 40 and 59 years old, in those under 20 years 64,9 percent were not. They presented symptoms of the disease upon admission 53,5 percent were asymptomatic, 76,6 percent of the positive radiographs corresponded to the symptomatic ones, coughing was the most frequent symptom. The greatest positivity in the chest X-ray was found in patients older than 60 years, the most frequent pattern was the opacity in the peripheral distribution veil. Conclusions: Asymptomatic patients predominate, the positivity of radiographs is higher in the elderly(AU)

Current limitations to identify COVID-19 using artificial intelligence with chest X-ray imaging.

The scientific community has joined forces to mitigate the scope of the current COVID-19 pandemic. The early identification of the disease, as well as the evaluation of its evolution is a primary task for the timely application of medical protocols. The use of medical images of the chest provides valuable information to specialists. Specifically, chest X-ray images have been the focus of many investigations that apply artificial intelligence techniques for the automatic classification of this disease. The results achieved to date on the subject are promising. However, some results of these investigations contain errors that must be corrected to obtain appropriate models for clinical use. This research discusses some of the problems found in the current scientific literature on the application of artificial intelligence techniques in the automatic classification of COVID-19. It is evident that in most of the reviewed works an incorrect evaluation protocol is applied, which leads to overestimating the results.

Infecciones bacterianas asociadas a la COVID-19 en pacientes de una unidad de cuidados intensivos / Bacterial infections associated with COVID-19 in patients of the Intensive Care Unit

Introducción: Los primeros informes de China sugirieron que la coinfección con otros patógenos en la COVID-19 era anómala, las últimas evidencias han demostrado que pueden aparecer otras infecciones, sobre todo en pacientes graves. Objetivo: Describir las infecciones bacterianas asociadas a la COVID-19, en pacientes de una unidad de cuidados intensivos. Métodos: Se realizó un estudio descriptivo en el período comprendido de marzo 24 a mayo 24 del año 2020, en la unidad de cuidados intensivos del Hospital Militar "Comandante Manuel Fajardo Rivero". La población de estudio estuvo constituida por 13 pacientes de 49 a 91 años, quienes permanecieron hospitalizados en esa sala, con diagnóstico confirmado, por la prueba de reacción en cadena de la transcriptasa inversa - polimerasa en tiempo real, para el SARS-CoV-2. Las variables de estudio fueron: edad, sexo, confección, antecedentes patológicos personales, estado al egreso, microorganismos aislados y susceptibilidad antimicrobiana. Resultados: El 61,5 por ciento de los pacientes fueron del sexo femenino, la edad media fue de 78,8 años, el 61,5 por ciento falleció y entre estos, el 44,4 por ciento presentó coinfección. El 66,7 por ciento y el 55,6 por ciento de los que padecían hipertensión arterial y cardiopatía isquémica respectivamente, desarrollaron una coinfección. La Escherichia coli fue el microorganismo que se aisló con mayor frecuencia. Conclusiones: En la serie estudiada predominaron las féminas, la mortalidad fue alta, se evidenció un porcentaje elevado de confección bacteriana y de comorbilidades. Más de la mitad de los pacientes falleció. Fueron las bacterias gramnegativas los microorganismos que más se aislaron. Los niveles de resistencia a los antimicrobianos fueron elevados(AU)

Introduction: The first reports from China suggested that coinfection with other pathogens in COVID-19 was abnormal, the latest evidence has shown that other infections may appear, especially in severe patients. Objective: To describe the bacterial infections associated with COVID-19, in patients in an intensive care unit. Methods: A descriptive study was carried out in the period from March 24 to May 24, 2020, in the intensive care unit of the Military Hospital "Comandante Manuel Fajardo Rivero". The study population consisted of 13 patients from 49 to 91 years, those who remained hospitalized in that room, with a confirmed diagnosis, by the real-time reverse transcriptase-polymerase chain reaction test for SARS-CoV-2. The study variables were: age, sex, clothing, personal pathological history, status at discharge, isolated microorganisms and antimicrobial susceptibility. Results: 61.5 percent of the patients were female, the mean age was 78.8 years, 61.5 percent died, and among these, 44.4 percent had coinfection. 66.7 percent and 55.6 percent of those with high blood pressure and ischemic heart disease, respectively, developed a coinfection. Escherichia coli was the most frequently isolated microorganism. Conclusions: Females predominated in the series studied, mortality was high, a high percentage of bacterial preparation and comorbidities was evident. More than half of the patients died. Gram-negative bacteria were the microorganisms that were most isolated. Antimicrobial resistance levels were high(AU)

Síndrome de Percheron / Percheron syndrome

El infarto talámicoparamediano bilateral sincrónico,llamado habitualmente infarto de la arteria de Percheron, se considera infrecuente y de difícil diagnóstico clínico. Se presenta a una paciente de 50 años, con infarto talámico bilateral, que presentó un cuadro de desorientación, visión borrosa y doble en horas de la mañana, sin referir pérdida de conciencia. En la tomografía axial computarizada de cráneo simple se evidenció una hipodensidad talámica bilateral, compatible con un infarto agudo a este nivel, por oclusión de la arteria de Percheron. Es importante el reconocimiento de esta variante anatómica para establecer el mecanismo del infarto talámico bilateral.

Síndrome de Percheron / Percheron syndrome

El infarto talámicoparamediano bilateral sincrónico, llamado habitualmente infarto de la arteriade Percheron, se considera infrecuente y de difícil diagnóstico clínico. Se presenta a una paciente de 50 años, con infarto talámico bilateral, que presentó un cuadro de desorientación, visión borrosa y doble en horas de la mañana, sin referir pérdida de conciencia. En la tomografía axialcomputarizada de cráneo simple se evidenció una hipodensidad talámica bilateral, compatible con un infarto agudo a este nivel, por oclusión de la arteria de Percheron. Es importante elreconocimiento de esta variante anatómica para establecer el mecanismo del infarto talámicobilateral(AU)