A new model for classification of medical CT images using CNN: a COVID-19 case study.

SARS-CoV-2 is the causative agent of COVID-19 and leaves characteristic impressions on chest Computed Tomography (CT) images in infected patients and this analysis is performed by radiologists through visual reading of lung images, and failures may occur. In this article, we propose a classification model, called Wavelet Convolutional Neural Network (WCNN) that aims to improve the differentiation of images of patients with COVID-19 from images of patients with other lung infections. The WCNN model was based on a Convolutional Neural Network (CNN) and wavelet transform. The model proposes a new input layer added to the neural network, which was called Wave layer. The hyperparameters values were defined by ablation tests. WCNN was applied to chest CT images to images from two internal and one external repositories. For all repositories, the average results of Accuracy (ACC), Sensitivity (Sen) and Specificity (Sp) were calculated. Subsequently, the average results of the repositories were consolidated, and the final values were ACC = 0.9819, Sen = 0.9783 and Sp = 0.98. The WCNN model uses a new Wave input layer, which standardizes the network input, without using data augmentation, resizing and segmentation techniques, maintaining the integrity of the tomographic image analysis. Thus, applications developed based on WCNN have the potential to assist radiologists with a second opinion in the analysis.1.

Nutritional state assessed by ultrasonography, but not by bioelectric impedance, predicts 28-day mortality in critically ill patients. Prospective cohort study.

BACKGROUND & AIMS: Malnutrition is a disease associated with unfavorable outcomes and high prevalence among critically ill patients. The primary aim of the study was to correlate the nutritional status of patients recently admitted to the intensive care unit (ICU) with the 28-day mortality rate. The secondary aim was to establish cut-off values for the bioelectrical impedance analysis (BIA) and ultrasonography (USG) methods for the classification of the nutritional status of critically ill patients. METHODS: Cohort-type prospective observational study design, with the inclusion of critically ill clinical-surgical patients, ≥18 years old, mechanically ventilated. Their nutritional status was assessed by phase angle and skeletal muscle mass (using BIA), and the sectional area of the rectus femoris muscle (RFM) (using USG) within the first 48 h of ICU admission. RESULTS: Among the patients included (n = 160), the non-survivors had lower values related to the survivors for averages of the phase angle (4.4 ± 1.5° vs 4.9 ± 1.2°; p = 0.041) and sectional area of the RFM (6.1 ± 1.6 cm2/m2vs 7.4 ± 1.8 cm2/m2; p = 0.030). Cut-off values were identified using the ROC curves for phase angle (≤4.4°), skeletal muscle mass (≤19.2 kg/m2), and sectional area of the RFM (≤5.9 cm2/m2). In multivariate logistic regression analysis, the sectional area of the RFM (≤5.9 cm2/m2) was the only independent predictor of 28-day mortality (OR = 6.08; p = 0.028). In the survival analysis, greater survival at 28 days was demonstrated for patients with phase angle >4.4° (p = 0.020) and sectional area of the RFM >5.9 cm2/m2 (p = 0.001). Malnourished patients had higher mortality than nourished patients according to the phase angle (31.5% vs 15.1%, respectively; p = 0.015) and the sectional area of the RFM (38.9% vs 6.3%, respectively; p = 0.001). CONCLUSIONS: The USG of the RFM is an efficient method for both classifying the nutritional status at ICU admission and assessing the prognosis of critically ill patients.

Novel Mahalanobis-based feature selection improves one-class classification of early hepatocellular carcinoma.

Detection of early hepatocellular carcinoma (HCC) is responsible for increasing survival rates in up to 40%. One-class classifiers can be used for modeling early HCC in multidetector computed tomography (MDCT), but demand the specific knowledge pertaining to the set of features that best describes the target class. Although the literature outlines several features for characterizing liver lesions, it is unclear which is most relevant for describing early HCC. In this paper, we introduce an unconstrained GA feature selection algorithm based on a multi-objective Mahalanobis fitness function to improve the classification performance for early HCC. We compared our approach to a constrained Mahalanobis function and two other unconstrained functions using Welch's t-test and Gaussian Data Descriptors. The performance of each fitness function was evaluated by cross-validating a one-class SVM. The results show that the proposed multi-objective Mahalanobis fitness function is capable of significantly reducing data dimensionality (96.4%) and improving one-class classification of early HCC (0.84 AUC). Furthermore, the results provide strong evidence that intensity features extracted at the arterial to portal and arterial to equilibrium phases are important for classifying early HCC.

Artifacts and pitfalls in shoulder magnetic resonance imaging.

Magnetic resonance imaging has revolutionized the diagnosis of shoulder lesions, in many cases becoming the method of choice. However, anatomical variations, artifacts and the particularity of the method may be a source of pitfalls, especially for less experienced radiologists. In order to avoid false-positive and false-negative results, the authors carried out a compilation of imaging findings that may simulate injury. It is the authors' intention to provide a useful, consistent and comprehensive reference for both beginner residents and skilled radiologists who work with musculoskeletal magnetic resonance imaging, allowing for them to develop more precise reports and helping them to avoid making mistakes.

A ressonância magnética revolucionou o diagnóstico de lesões do ombro, tornando-se, em muitos casos, o método de escolha. No entanto, as variações anatômicas, artefatos e particularidade do método podem ser fonte de armadilhas, especialmente para radiologistas menos experientes. Para evitar resultados falso-positivos e falso-negativos, foi realizada uma compilação de achados de imagem que podem simular lesões. Pretendemos ser uma referência útil, consistente e abrangente para os residentes iniciantes e radiologistas qualificados que trabalham com ressonância magnética musculoesquelética, a fim de desenvolver relatórios mais precisos e ajudá-los a evitar erros.

Artifacts and pitfalls in shoulder magnetic resonance imaging / Artefatos e armadilhas na ressonância magnética do ombro

AbstractMagnetic resonance imaging has revolutionized the diagnosis of shoulder lesions, in many cases becoming the method of choice. However, anatomical variations, artifacts and the particularity of the method may be a source of pitfalls, especially for less experienced radiologists. In order to avoid false-positive and false-negative results, the authors carried out a compilation of imaging findings that may simulate injury. It is the authors’ intention to provide a useful, consistent and comprehensive reference for both beginner residents and skilled radiologists who work with musculoskeletal magnetic resonance imaging, allowing for them to develop more precise reports and helping them to avoid making mistakes.

ResumoA ressonância magnética revolucionou o diagnóstico de lesões do ombro, tornando-se, em muitos casos, o método de escolha. No entanto, as variações anatômicas, artefatos e particularidade do método podem ser fonte de armadilhas, especialmente para radiologistas menos experientes. Para evitar resultados falso-positivos e falso-negativos, foi realizada uma compilação de achados de imagem que podem simular lesões. Pretendemos ser uma referência útil, consistente e abrangente para os residentes iniciantes e radiologistas qualificados que trabalham com ressonância magnética musculoesquelética, a fim de desenvolver relatórios mais precisos e ajudá-los a evitar erros.

Distinção entre os tipos 1 e 2 de tireotoxicose associada à amiodarona por meio de dúplex-doppler colorido / Differentiation between types 1 and 2 of amiodarone-associated thyrotoxicosis using color duplex sonography

INTRODUÇÃO: A amiodarona pode causar tireotoxicose, principalmente em áreas geográficas onde a ingestão de iodo é insuficiente. Dois tipos distintos de tireotoxicose associada à amiodarona (TAA) podem ser encontrados: A) o tipo 1 - a doença é secundária à sobrecarga de iodo (fenômeno Jod-Basedow), geralmente encontrada em indivíduos com doença tireóidea preexistente, B) o tipo 2 - quando a tireotoxicose deve-se a uma tireoidite destrutiva, com ruptura folicular e liberação do conteúdo folicular. A distinção entre os dois tipos é fundamental para a conduta terapêutica. Este estudo transversal, realizado no Instituto de Radiologia do Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo entre janeiro de 2004 a março de 2006, objetivou: A) Demonstrar a utilidade da densidade de pixéis coloridos (DPC), parâmetro objetivo obtido por meio de programa de computador, na distinção entre os dois tipos de TAA. B) Determinar os melhores critérios objetivos de distinção entre os dois tipos de TAA por meio da ultra-sonografia dúplex-Doppler colorido da tireóide. C) Conhecer o grau de concordância intra-observadores e interobservadores dos padrões subjetivos de vascularização do parênquima tireóideo. MÉTODOS: Foram examinadas 158 glândulas tireóideas por meio de dúplex-Doppler colorido. Após passagem pelos critérios de seleção, 137 indivíduos foram divididos em quatro grupos: Grupo N (n = 84), grupo A (n = 30), grupo I (n = 14) e grupo II (n = 9), compostos por indivíduos normais, eutireóideos em uso de amiodarona, pacientes com tireotoxicose tipos 1 e 2, respectivamente. Todos os indivíduos foram submetidos ao estudo dúplex-Doppler colorido da tireóide e testes laboratoriais. Os pacientes em tireotoxicose também realizaram captação de iodo radioativo em 24 horas. RESULTADOS: No grupo I, tanto a densidade de pixéis coloridos (DPC = 17,22 ± 20,81%) quanto as velocidades de pico sistólico nas artérias tireóideas superiores (VSTS = 38,54 ± 18,62 cm/s)...

INTRODUCTION: Amiodarone can cause thyrotoxicosis, mainly in geographic insufficient iodine intake areas. Two different types of amiodarone-associated thyrotoxicosis can be found: A) type 1 - it occurs in patients with preexistent thyroid disease, such as goiter or autonomous nodule, and the iodine load associated with amiodarone triggers increased synthesis of thyroid hormones (Jod-Basedow fenomenon), B) type 2 - a destructive thyroiditis is found, characterized by follicular rupture and release of its content. In order to treat appropriately, the differentiation between the two types is crucial. This transversal study, carried through at the Institute of Radiology, Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo from January 2004 to March 2006, aimed to: A) demonstrate the utility of the color pixel density (CPD), a semiquantitative parameter gotten by means of computer program, in the differentiation between the two types of AAT, B) verify the best objective criteria of distinction between the two types of AAT through the duplex-Doppler sonography of the thyroid, C) investigate the agreement intra and interobservers of the subjective classification of the thyroid vascularization. METHODS: thyroid glands of 158 subjects were examined by means of duplex-Doppler sonography. After all the selection criteria, 137 individuals were selected into four groups; group N (n = 84), group A (n = 30), group I (n = 14) and II (n = 9), compounded of normal individuals, euthyroids in use of amiodarone, and patients with thyrotoxicosis types 1 and 2, respectively. All the individuals were submitted to perform the duplex-Doppler sonography of the thyroid and laboratorial tests. Also, the patients with thyrotoxicosis had also carried through 24 hour radioactive iodine uptake. RESULTS: In the group I, not only the color pixel density (CPD = 17.22 ± 20.81%) but also the values of systolic peak velocity in the superior and the inferior thyroid...

[Ophthalmoplegic migraine: MRI findings. Case report]. / Enxaqueca oftalmoplégica: achados à ressonância magnética. Relato de caso.

Ophthalmoplegic migraine is a rare syndrome in which headache is associated with ophthalmoplegia and third, fourth or sixth cranial nerves palsy. It occurs most frequently in childhood and teenagers. At magnetic resonance imaging (MRI) with gadolinium (GD-DTPA) it may be observed a transitory enhancement of the affected nerve. We present the case of a male teenager, 16 years old, with typical medical history and enhanced signal at left oculomotor nerve in cisternal portion at MRI weighted in T1 with GD-DTPA. On the control exam, eighteen months later, there was no remarkable lesion. The enhancement of oculomotor nerve at MRI is always pathological and among the differential diagnosis we must include: neoplasia (lymphoma and leukemia), infections (AIDS, syphilis), inflammatory process (sarcoidose and Tolosa-Hunt syndrome) and vascular (posterior communicating artery aneurysm).

Histiocitoma fibroso benigno profundo: achados na tomografia computadorizada e histologia / Deep benign fibrous histiocytoma: computed tomography and histology findings

Apresentamos imagens de tomografia computadorizada de um paciente de 83 anos de idade com histiocitoma fibroso benigno profundo, na face lateral da perna esquerda, com evolução de longa data. A tomografia computadorizada observamos massa bem delimitada, ovalada, e que apresentava intenso realce periférico com uso do meio de contraste iodado. Este é um tumor de partes moles pouco freqüente e com raros relatos dos seus aspectos de imagem.

Enxaqueca oftalmoplégica: achados à ressonância magnética. Relato de caso / Ophthalmoplegic migraine: MRI fingings. Case report

A enxaqueca oftalmoplégica é síndrome rara em que a cefaléia se associa à oftalmoplegia por paralisia do terceiro, quarto ou sexto nervos cranianos. Ocorre mais freqüentemente em crianças e adolescentes. Ao exame de ressonância magnética (RM), com uso de gadolínio, é observado realce transitório do nervo afetado. Apresentamos o caso de um adolescente, 16 anos, masculino, com história típica e hipersinal nas seqüências ponderadas em T1 pós-infusão venosa do gadolínio na porção cisternal do nervo oculomotor esquerdo. No exame de controle, realizado 18 meses depois, houve remissão completa das alterações na RM. O hipersinal do oculomotor à RM é sempre patológico; entre os diagnósticos diferenciais devemos incluir neoplasias (linfoma, leucemia), infecções (SIDA, sífilis), doenças inflamatórias não-infecciosas (sarcoidose, síndrome de Tolosa-Hunt) e vasculares (aneurisma da artéria comunicante posterior).

Síndrome de Hallervorden Spatz: achados na resonância magnética. Relato de caso / Hallervorden Spatz syndrome: magnetic resonance findings. Case report

A síndrome de Hallervorden-Spatz é afecção neurodegenerativa, autossômica recessiva com duas apresentações clínicas: precoce e tardia. Esta última é caracterizada pelo acometimento psiquiátrico e a presença de sinais piramidais e extrapiramidais. Relatamos o caso de mulher de 41 anos, com história de alterações extrapiramidais. O exame de ressonância magnética (RM) mostrou o sinal dos olhos-de-tigre, lesão dos globos pálidos mediais com deposição de ferro periférica (hipo-sinal) e gliose central (hipersinal), nas seqüências com tempo de repetição (TR) longo. Há forte relação entre o sinal na RM e mutações no gene responsável pela lesão, fazendo deste exame sensível o suficiente para a realização do diagnóstico da doença.

[Hallervorden Spatz syndrome: magnetic resonance findings. Case report]. / Síndrome de Hallervorden Spatz: achados na resonância magnética. Relato de caso.

Hallervorden-Spatz syndrome is a neurodegenerative disease, autosomic recessive with two clinical features: early and late onset. Psychiatric, pyramidal and extrapyramidal signs are present in the late subtype. We report the case of a 41-old woman with extrapyramidal signs. Magnetic resonance imaging (MRI) showed the eye-of-the-tiger sign in the medial globus pallidus. This is due to a gliosis (increased signal) and accumulation of surrounding iron (decreased signal intensity) in long TR sequences. There is a strong relationship between MRI findings and the gene mutation responsable for this disease. It makes the MRI sensible for diagnosing this syndrome.