ABSTRACT The novel coronavirus (COVID-19) is a disease that mainly affects the lung tissue. The detection of lesions caused by this disease can help to provide an adequate treatment and monitoring its evolution. This research focuses on the bi- nary classification of lung lesions caused by COVID-19 in images of computed tomography (CT) using deep learning. The database used in the experiments comes from two independent repositories, which contains tomographic scans of patients with a positive diagnosis of COVID-19. The output layers of four pre-trained convolutional networks were adapted to the proposed task and re-trained using the fine-tuning technique. The models were validated with test images from the two database’s repositories. The model VGG19, considering one of the repositories, showed the best performance with 88% and 90.2% of accuracy and recall, respectively. The model combination using the soft voting technique presented the highest accuracy (84.4%), with a recall of 94.4% employing the data from the other repository. The area under the receiver operating characteristic curve was 0.92 at best. The proposed method based on deep learning represents a valuable tool to automatically classify COVID-19 lesions on CT images and could also be used to assess the extent of lung infection.

Abstract: Positron Emission Tomography (PET) is a nuclear medicine technique used to obtain metabolic images of the body. PET scanners used in the research, treatment, and monitoring of several diseases provide images of metabolic activity associated with the ailments. However, the data produced by PET are heavily corrupted by noise and other errors, thereby causing degradation in the quality of the final reconstructed images. In order to improve the image reconstruction process, this paper presents a new algorithm that addresses the problem from a variational perspective. We propose the use of a modified version of total variation regularization by including a second term in order to better deal with noise; in the proposed version, both regularizing terms are balanced by calculating weights adapted to the PET images through the use of anatomical information from another medical modality, such as computer tomography (CT) or magnetic resonance imaging (MRI). Simulated image results show that our proposed method is more effective in dealing with heavy noise and in preserving small structures (e.g., possible lesions) than the expectation maximization method that is commonly used with commercial scanners.

Postsurgical pseudoaneurysms of the ascending aorta are a rare and serious complication of cardiovascular surgery that requires a surgical solution of urgency due to its high complexity. A case of a 40 year old female, white skin color, that due to severe aortic regurgitation underwent surgical treatment for aortic valve replacement. During the immediate postoperative period began with a fever, he made a meticulous study that included contrast computed tomography, through which diagnosed the presence of a pseudoaneurysm of the ascending aorta, was taken to the operating room, they found two sacks pseudoaneurismatics at the site where the suture was carried out previous.

Los pseudoaneurismas postquirúrgicos ascendentes de la aórtica, son una rara y grave complicación de la cirugía cardiovascular que requiere de una solución quirúrgica de máxima urgencia debido a su alta complejidad. Se presenta el caso de una paciente de 40 años de edad, de sexo femenino, de color de piel blanca, que producto de una insuficiencia aórtica severa se sometió a tratamiento quirúrgico, para realizar una sustitución valvular aórtica. Durante el periodo postoperatorio mediato comenzó con un cuadro febril, se le realizó un estudio minucioso que incluyó tomografía axial computarizada contrastada, por medio de la cual se diagnosticó la presencia de un pseudoaneurisma de la aorta ascendente, fue llevada al quirófano, se encontraron dos sacos pseudoaneurismáticos en el sitio donde se realizó de la aortorrafía anterior.

We describe a new method for surface reconstruction and smoothing based on unorganized noisy point clouds without normals. The output of the method is a refined triangular mesh that approximates the original point cloud while preserving the salient features of the underlying surface. The method has five steps: noise removal, clustering, data reduction, initial reconstruction, and mesh refinement. We also present theoretical justifications for the heuristics used in the reconstruction step.