A computing platform to analyze breast abnormalities using infrared images.

The present work shows a computational tool developed in the MATLAB platform. Its main functionality is to evaluate a thermal model of the breast. This computational infrastructure consists of modules in which manipulate the infrared images and calculate breast temperature profiles. It also allows the analysis of breast nodules. The different modules of the framework are interconnected through an interface which the major purpose is to automatize the whole process of the infrared image analysis, in a quick and organized way. The tool is initially supplied with a three-dimensional mesh that represents the substitute geometry of the patient's breast together with her infrared images which are transformed into temperature matrices. Through these matrices, the frontal and lateral mappings are performed by specified modules. This process generates an image and a text file with all the temperatures associated to the nodes of the surface mesh. The developed tool is also able to manage the use of a commercial mesh generation program and a computational fluid dynamics code, the FLUENT, in order to validate the technique by the use of a parametric analysis. In these analyses, the tumor may have several geometric shapes and different locations within the breast.

Dental arch definition in computed tomographs using two semi-automatic methods.

Computed tomography is a widely used image examination in dental imaging that provides an accurate location of oral structures and features, including the dental arch, which is an important anatomical feature. This study proposes two new semi-automatic methods for arch definition in CTs, with minimal user effort. This study includes 25 CT examinations. The first method is based on the teeth pulps, and the second one is based on the whole mandible. The methods use thresholding and morphological operations to obtain the arches. The evaluation process includes two different metrics DTW and IoU. For both metrics, the initial results of M1 were very low, but the average performance of M2 can be considered high. The analysis showed that changing the input improves the M1 results substantially. The promising results presented here suggest that these methods can be used as auxiliary tools for the proposed task.

Multivariate data driven prediction of COVID-19 dynamics: Towards new results with temperature, humidity and air quality data.

Since the start of the COVID-19 pandemic many studies investigated the correlation between climate variables such as air quality, humidity and temperature and the lethality of COVID-19 around the world. In this work we investigate the use of climate variables, as additional features to train a data-driven multivariate forecast model to predict the short-term expected number of COVID-19 deaths in Brazilian states and major cities. The main idea is that by adding these climate features as inputs to the training of data-driven models, the predictive performance improves when compared to equivalent single input models. We use a Stacked LSTM as the network architecture for both the multivariate and univariate model. We compare both approaches by training forecast models for the COVID-19 deaths time series of the city of São Paulo. In addition, we present a previous analysis based on grouping K-means on AQI curves. The results produced will allow achieving the application of transfer learning, once a locality is eventually added to the task, regressing out using a model based on the cluster of similarities in the AQI curve. The experiments show that the best multivariate model is more skilled than the best standard data-driven univariate model that we could find, using as evaluation metrics the average fitting error, average forecast error, and the profile of the accumulated deaths for the forecast. These results show that by adding more useful features as input to a multivariate approach could further improve the quality of the prediction models.

Modeling the basic behaviors of Anesthesia Training in Relation to Puncture and Penetration Feedback.

Failure rates in spinal anesthesia are generally low in experienced hands. However, studies report a failure rate variation of 1% to 17% in this procedure. The aim of this study is to bring the main characteristics of in vivo procedure to the virtual reality simulated environment. The first step is to model the behavior of tissue layers being punctured by a needle to then make its inclusion in medical training possible. The simulation proposed here is implemented using a Phantom Omni haptic device. Every crucial sensation of the method mentioned here was assessed by a dozen volunteers who participated in two experiments designed to validate the modeled response. Each user answered six questions (three for each experiment). Good results were achieved in certain essential aspects of the process, such as identifying the number of layers, the most rigid layer to puncture, and the most resistant layers to pass through. These results indicated that it is possible to represent many typical behaviors through virtual needle insertion in spinal anesthesia with the correct use of haptic properties.Clinical relevance- The idea is to create a spinal anesthesia simulator that could work as a complementary step in training new anesthetists. The use of a simulator avoids introducing the first puncture haptic sensation directly in patients.

Classification of Approximal Caries in Bitewing Radiographs Using Convolutional Neural Networks.

Dental caries is an extremely common problem in dentistry that affects a significant part of the population. Approximal caries are especially difficult to identify because their position makes clinical analysis difficult. Radiographic evaluation-more specifically, bitewing images-are mostly used in such cases. However, incorrect interpretations may interfere with the diagnostic process. To aid dentists in caries evaluation, computational methods and tools can be used. In this work, we propose a new method that combines image processing techniques and convolutional neural networks to identify approximal dental caries in bitewing radiographic images and classify them according to lesion severity. For this study, we acquired 112 bitewing radiographs. From these exams, we extracted individual tooth images from each exam, applied a data augmentation process, and used the resulting images to train CNN classification models. The tooth images were previously labeled by experts to denote the defined classes. We evaluated classification models based on the Inception and ResNet architectures using three different learning rates: 0.1, 0.01, and 0.001. The training process included 2000 iterations, and the best results were achieved by the Inception model with a 0.001 learning rate, whose accuracy on the test set was 73.3%. The results can be considered promising and suggest that the proposed method could be used to assist dentists in the evaluation of bitewing images, and the definition of lesion severity and appropriate treatments.

Pulmonary COVID-19: Learning Spatiotemporal Features Combining CNN and LSTM Networks for Lung Ultrasound Video Classification.

Deep Learning is a very active and important area for building Computer-Aided Diagnosis (CAD) applications. This work aims to present a hybrid model to classify lung ultrasound (LUS) videos captured by convex transducers to diagnose COVID-19. A Convolutional Neural Network (CNN) performed the extraction of spatial features, and the temporal dependence was learned using a Long Short-Term Memory (LSTM). Different types of convolutional architectures were used for feature extraction. The hybrid model (CNN-LSTM) hyperparameters were optimized using the Optuna framework. The best hybrid model was composed of an Xception pre-trained on ImageNet and an LSTM containing 512 units, configured with a dropout rate of 0.4, two fully connected layers containing 1024 neurons each, and a sequence of 20 frames in the input layer (20×2018). The model presented an average accuracy of 93% and sensitivity of 97% for COVID-19, outperforming models based purely on spatial approaches. Furthermore, feature extraction using transfer learning with models pre-trained on ImageNet provided comparable results to models pre-trained on LUS images. The results corroborate with other studies showing that this model for LUS classification can be an important tool in the fight against COVID-19 and other lung diseases.

A Computational Study on the Role of Parameters for Identification of Thyroid Nodules by Infrared Images (and Comparison with Real Data).

According to experts and medical literature, healthy thyroids and thyroids containing benign nodules tend to be less inflamed and less active than those with malignant nodules. It seems to be a consensus that malignant nodules have more blood veins and more blood circulation. This may be related to the maintenance of the nodule's heat at a higher level compared with neighboring tissues. If the internal heat modifies the skin radiation, then it could be detected by infrared sensors. The goal of this work is the investigation of the factors that allow this detection, and the possible relation with any pattern referent to nodule malignancy. We aim to consider a wide range of factors, so a great number of numerical simulations of the heat transfer in the region under analysis, based on the Finite Element method, are performed to study the influence of each nodule and patient characteristics on the infrared sensor acquisition. To do so, the protocol for infrared thyroid examination used in our university's hospital is simulated in the numerical study. This protocol presents two phases. In the first one, the body under observation is in steady state. In the second one, it is submitted to thermal stress (transient state). Both are simulated in order to verify if it is possible (by infrared sensors) to identify different behavior referent to malignant nodules. Moreover, when the simulation indicates possible important aspects, patients with and without similar characteristics are examined to confirm such influences. The results show that the tissues between skin and thyroid, as well as the nodule size, have an influence on superficial temperatures. Other thermal parameters of thyroid nodules show little influence on surface infrared emissions, for instance, those related to the vascularization of the nodule. All details of the physical parameters used in the simulations, characteristics of the real nodules and thermal examinations are publicly available, allowing these simulations to be compared with other types of heat transfer solutions and infrared examination protocols. Among the main contributions of this work, we highlight the simulation of the possible range of parameters, and definition of the simulation approach for mapping the used infrared protocol, promoting the investigation of a possible relation between the heat transfer process and the data obtained by infrared acquisitions.

A hybrid methodology for breast screening and cancer diagnosis using thermography.

Breast cancer is the second most common cancer in the world. Early diagnosis and treatment increase the patient's chances of healing. The temperature of cancerous tissues is generally different from that of healthy neighboring tissues, making thermography an option to be considered in the fight against cancer because it does not use ionizing radiation, venous access, or any other invasive process, presenting no damage or risk to the patient. In this paper, we propose a hybrid computational method using the Dynamic Infrared Thermography (DIT) and Static Infrared Thermography (SIT) for abnormality screening and diagnosis of malignant tumor (cancer), applying supervised and unsupervised machine learning techniques. We use the area under receiver operating characteristic curve, sensitivity, specificity, and accuracy as performance measures to compare the hybrid methodology with previous work in the literature. The K-Star classifier achieved accuracy of 99% in the screening phase using DIT images. The Support Vector Machines (SVM) classifier applied on SIT images yielded accuracy of 95% in the diagnosis of cancer. The results confirm the potential of the proposed approaches for screening and diagnosis of breast cancer.

Combining Genetic Algorithms and SVM for Breast Cancer Diagnosis Using Infrared Thermography.

Breast cancer is one of the leading causes of mortality globally, but early diagnosis and treatment can increase the cancer survival rate. In this context, thermography is a suitable approach to help early diagnosis due to the temperature difference between cancerous tissues and healthy neighboring tissues. This work proposes an ensemble method for selecting models and features by combining a Genetic Algorithm (GA) and the Support Vector Machine (SVM) classifier to diagnose breast cancer. Our evaluation demonstrates that the approach presents a significant contribution to the early diagnosis of breast cancer, presenting results with 94.79% Area Under the Receiver Operating Characteristic Curve and 97.18% of Accuracy.

Do Radiographic Assessments of Periodontal Bone Loss Improve with Deep Learning Methods for Enhanced Image Resolution?

Resolution plays an essential role in oral imaging for periodontal disease assessment. Nevertheless, due to limitations in acquisition tools, a considerable number of oral examinations have low resolution, making the evaluation of this kind of lesion difficult. Recently, the use of deep-learning methods for image resolution improvement has seen an increase in the literature. In this work, we performed two studies to evaluate the effects of using different resolution improvement methods (nearest, bilinear, bicubic, Lanczos, SRCNN, and SRGAN). In the first one, specialized dentists visually analyzed the quality of images treated with these techniques. In the second study, we used those methods as different pre-processing steps for inputs of convolutional neural network (CNN) classifiers (Inception and ResNet) and evaluated whether this process leads to better results. The deep-learning methods lead to a substantial improvement in the visual quality of images but do not necessarily promote better classifier performance.

Hyperparameter Optimization for COVID-19 Pneumonia Diagnosis Based on Chest CT.

Convolutional Neural Networks (CNNs) have been successfully applied in the medical diagnosis of different types of diseases. However, selecting the architecture and the best set of hyperparameters among the possible combinations can be a significant challenge. The purpose of this work is to investigate the use of the Hyperband optimization algorithm in the process of optimizing a CNN applied to the diagnosis of SARS-Cov2 disease (COVID-19). The test was performed with the Optuna framework, and the optimization process aimed to optimize four hyperparameters: (1) backbone architecture, (2) the number of inception modules, (3) the number of neurons in the fully connected layers, and (4) the learning rate. CNNs were trained on 2175 computed tomography (CT) images. The CNN that was proposed by the optimization process was a VGG16 with five inception modules, 128 neurons in the two fully connected layers, and a learning rate of 0.0027. The proposed method achieved a sensitivity, precision, and accuracy of 97%, 82%, and 88%, outperforming the sensitivity of the Real-Time Polymerase Chain Reaction (RT-PCR) tests (53-88%) and the accuracy of the diagnosis performed by human experts (72%).

Application of thermography in the diagnostic investigation of thyroid nodules.

Thyroid nodules (TN) are common in the general population, and the clinical importance of diagnosing thyroid nodules is based on excluding the possibility of thyroid cancer, which occurs in 7-15% of cases. The thyroid gland, owing to its superficial location, is easily accessible via thermography, a noninvasive method of recording body temperature that measures infrared radiation emitted by the body surface. Therefore, this study aimed to evaluate the temperature differences between benign and malignant TN by using thermography. We conducted a cross-sectional study where 147 TN were divided into two groups: the first group included 120 benign nodules and the other included 27 malignant nodules. All the nodules were subjected to ultrasound, fine needle aspiration biopsy, and thermography. On analyzing the thermography results, the benign nodules had a higher temperature at the beginning of the thermography evaluation, and the malignant nodules showed a higher temperature in the middle and at the end (Ft). Using the relationships, it was observed that the temperature delta (ΔT), ΔT nodule/ΔT healthy, ΔT nodule minus ΔT healthy, and nodule Ft minus Ft of the healthy region were higher in malignant nodules. The ROC curve analysis of ΔT demonstrated a cutoff point of 2.38°C, with a sensitivity of 0.963 and specificity of 0.992. Malignant nodules have higher temperatures than benign nodules on thermographic evaluation. This finding suggests that thermography can be a useful tool in the diagnosis of thyroid nodules.

Using super-resolution generative adversarial network models and transfer learning to obtain high resolution digital periapical radiographs.

Periapical Radiographs are commonly used to detect several anomalies, like caries, periodontal, and periapical diseases. Even considering that digital imaging systems used nowadays tend to provide high-quality images, external factors, or even system limitations can result in a vast amount of radiographic images with low quality and resolution. Commercial solutions offer tools based on interpolation methods to increase image resolution. However, previous literature shows that these methods may create undesirable effects in the images affecting the diagnosis accuracy. One alternative is using deep learning-based super-resolution methods to achieve better high-resolution images. Nevertheless, the amount of data for training such models is limited, demanding transfer learning approaches. In this work, we propose the use of super-resolution generative adversarial network (SRGAN) models and transfer learning to achieve periapical images with higher quality and resolution. Moreover, we evaluate the influence of using the transfer learning approach and the datasets selected for it in the final generated images. For that, we performed an experiment comparing the performance of the SRGAN models (with and without transfer learning) with other super-resolution methods. Considering Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Mean Opinion Score (MOS), the results of SRGAN models using transfer learning were better on average. This superiority was also verified statistically using the Wilcoxon paired test. In the visual analysis, the high quality achieved by the SRGAN models, in general, is visible, resulting in more defined edges details and fewer blur effects.

A dynamic decision model for diagnosis of dementia, Alzheimer's disease and Mild Cognitive Impairment.

CDSS (Clinical Decision Support System) is a domain within digital health that aims at supporting clinicians by suggesting the most probable diagnosis based on knowledge obtained from patient data. Usually, decision models used by current CDSS are static, i.e., they are not updated when new data are included, which could allow them to acquire new knowledge and enhance system accuracy. This paper proposes a dynamic decision model that automatically updates itself from classifier models using supervised machine learning algorithms. Our supervised learning process ranks several decision models using classifier performance measures, considering available patient data, filled by the health center, or local clinical guidelines. The decision model with the best performance is then selected to be used in our CDSS, which is designed for the diagnosis of D (Dementia), AD (Alzheimer's Disease), and MCI (Mild Cognitive Impairment). Patient datasets from CAD (Center for Alzheimer's Disease), at the Institute of Psychiatry of UFRJ (Federal University of Rio de Janeiro), and CRASI (Center of Reference in Attention to Health of the Elderly), at Antonio Pedro Hospital of UFF (Fluminense Federal University), are used. The main conclusion is that the proposed dynamic decision model, which offers the ability to be continuously refined with more recent diagnostic criteria or even personalized according to the local domain or clinical guidelines, provides an efficient alternative for diagnosis of Dementia, AD, and MCI.

ELEMENT: Multi-Modal Retinal Vessel Segmentation Based on a Coupled Region Growing and Machine Learning Approach.

Vascular structures in the retina contain important information for the detection and analysis of ocular diseases, including age-related macular degeneration, diabetic retinopathy and glaucoma. Commonly used modalities in diagnosis of these diseases are fundus photography, scanning laser ophthalmoscope (SLO) and fluorescein angiography (FA). Typically, retinal vessel segmentation is carried out either manually or interactively, which makes it time consuming and prone to human errors. In this research, we propose a new multi-modal framework for vessel segmentation called ELEMENT (vEsseL sEgmentation using Machine lEarning and coNnecTivity). This framework consists of feature extraction and pixel-based classification using region growing and machine learning. The proposed features capture complementary evidence based on grey level and vessel connectivity properties. The latter information is seamlessly propagated through the pixels at the classification phase. ELEMENT reduces inconsistencies and speeds up the segmentation throughput. We analyze and compare the performance of the proposed approach against state-of-the-art vessel segmentation algorithms in three major groups of experiments, for each of the ocular modalities. Our method produced higher overall performance, with an overall accuracy of 97.40%, compared to 25 of the 26 state-of-the-art approaches, including six works based on deep learning, evaluated on the widely known DRIVE fundus image dataset. In the case of the STARE, CHASE-DB, VAMPIRE FA, IOSTAR SLO and RC-SLO datasets, the proposed framework outperformed all of the state-of-the-art methods with accuracies of 98.27%, 97.78%, 98.34%, 98.04% and 98.35%, respectively.

A Computational Method to Assist the Diagnosis of Breast Disease Using Dynamic Thermography.

Breast cancer has been the second leading cause of cancer death among women. New techniques to enhance early diagnosis are very important to improve cure rates. This paper proposes and evaluates an image analysis method to automatically detect patients with breast benign and malignant changes (tumors). Such method explores the difference of Dynamic Infrared Thermography (DIT) patterns observed in patients' skin. After obtaining the sequential DIT images of each patient, their temperature arrays are computed and new images in gray scale are generated. Then the regions of interest (ROIs) of those images are segmented and, from them, arrays of the ROI temperature are computed. Features are extracted from the arrays, such as the ones based on statistical, clustering, histogram comparison, fractal geometry, diversity indices and spatial statistics. Time series that are broken down into subsets of different cardinalities are generated from such features. Automatic feature selection methods are applied and used in the Support Vector Machine (SVM) classifier. In our tests, using a dataset of 68 images, 100% accuracy was achieved.

The role of malignant tissue on the thermal distribution of cancerous breast.

The present work focuses on the integration of analytical and numerical strategies to investigate the thermal distribution of cancerous breasts. Coupled stationary bioheat transfer equations are considered for the glandular and heterogeneous tumor regions, which are characterized by different thermophysical properties. The cross-section of the cancerous breast is identified by a homogeneous glandular tissue that surrounds the heterogeneous tumor tissue, which is assumed to be a two-phase periodic composite with non-overlapping circular inclusions and a square lattice distribution, wherein the constituents exhibit isotropic thermal conductivity behavior. Asymptotic periodic homogenization method is used to find the effective properties in the heterogeneous region. The tissue effective thermal conductivities are computed analytically and then used in the homogenized model, which is solved numerically. Results are compared with appropriate experimental data reported in the literature. In particular, the tissue scale temperature profile agrees with experimental observations. Moreover, as a novelty result we find that the tumor volume fraction in the heterogeneous zone influences the breast surface temperature.

An Infrared Thermal Images Database and a New Technique for Thyroid Nodules Analysis.

Thyroid nodules diseases are a common health problem and thyroidal cancer is becoming increasingly prevalent. They appear in the neck and bottom neck region, superficially over the trachea. Cancer tissues are characterized by higher temperatures than surrounding tissues. Thermography is a diagnostic tool increasingly used to detect cancer and abnormalities. Artificial intelligence is an approach which can be applied to thyroid nodules classification, but is necessary to have a proper number of cases with proven diagnosis. In this paper, a new database that contain infrared thermal images, clinical and physiological data is presented. The description of each nodule per patient, and the acquisition protocol (based on Dynamic Infrared Thermography approach) is considered as well. A semi-automatic method for image registration was implemented to pre-process the thermograms and a new method for the Region of Interest (ROI) extraction is proposed. Moreover, the obtained ROI results are confirmed by medical specialists and turned available for future comparison with other works.

Hybrid analysis for indicating patients with breast cancer using temperature time series.

Breast cancer is the most common cancer among women worldwide. Diagnosis and treatment in early stages increase cure chances. The temperature of cancerous tissue is generally higher than that of healthy surrounding tissues, making thermography an option to be considered in screening strategies of this cancer type. This paper proposes a hybrid methodology for analyzing dynamic infrared thermography in order to indicate patients with risk of breast cancer, using unsupervised and supervised machine learning techniques, which characterizes the methodology as hybrid. The dynamic infrared thermography monitors or quantitatively measures temperature changes on the examined surface, after a thermal stress. In the dynamic infrared thermography execution, a sequence of breast thermograms is generated. In the proposed methodology, this sequence is processed and analyzed by several techniques. First, the region of the breasts is segmented and the thermograms of the sequence are registered. Then, temperature time series are built and the k-means algorithm is applied on these series using various values of k. Clustering formed by k-means algorithm, for each k value, is evaluated using clustering validation indices, generating values treated as features in the classification model construction step. A data mining tool was used to solve the combined algorithm selection and hyperparameter optimization (CASH) problem in classification tasks. Besides the classification algorithm recommended by the data mining tool, classifiers based on Bayesian networks, neural networks, decision rules and decision tree were executed on the data set used for evaluation. Test results support that the proposed analysis methodology is able to indicate patients with breast cancer. Among 39 tested classification algorithms, K-Star and Bayes Net presented 100% classification accuracy. Furthermore, among the Bayes Net, multi-layer perceptron, decision table and random forest classification algorithms, an average accuracy of 95.38% was obtained.

On the Automated Segmentation of Epicardial and Mediastinal Cardiac Adipose Tissues Using Classification Algorithms.

The quantification of fat depots on the surroundings of the heart is an accurate procedure for evaluating health risk factors correlated with several diseases. However, this type of evaluation is not widely employed in clinical practice due to the required human workload. This work proposes a novel technique for the automatic segmentation of cardiac fat pads. The technique is based on applying classification algorithms to the segmentation of cardiac CT images. Furthermore, we extensively evaluate the performance of several algorithms on this task and discuss which provided better predictive models. Experimental results have shown that the mean accuracy for the classification of epicardial and mediastinal fats has been 98.4% with a mean true positive rate of 96.2%. On average, the Dice similarity index, regarding the segmented patients and the ground truth, was equal to 96.8%. Therfore, our technique has achieved the most accurate results for the automatic segmentation of cardiac fats, to date.