feature extraction

Breast cancer is one of the most common cancers in women and the main cause of death in cancer diseases, and mammography is the primary imaging method for early detection of breast masses. Rapid diagnosis with high accuracy is one of the serious concerns of doctors and healthcare centers when facing certain diseases, so the purpose of this article was to determine the automatic classification of BI-RADS in mammography reports using data fusion.

The present descriptive, analytical, and retrospective study was conducted in 2023, the mammography report and the electronic file of the patients were extracted from the archiving and communication system of the patient's image and records obtained from the available information in the medical training center of Shahidzadeh hospital in Behbahan, Iran, which includes the mammography reports and the electronic record of 250 patients who had ample information. To model the proposed method using the collected dataset, Python software was used in the Visual Studio Code environment. Finally, cross-validation was used to evaluate the quality and validity of the results.

The results confirmed that the proposed approach, namely the use of Word2vec combined with TFIDF, and their integration with HIS, had a significant impact on the accuracy of medical text classification. The output vectors of Word2vec were used for BI-RADS level classification when TFIDF was applied or not applied, as well as with and without the integration of HIS, for classifiers such as CNN, MLP, DT, and k-NN, and the results were compared using evaluation measures such as accuracy, precision, sensitivity, positive predictive value, negative predictive value, and F1 score. The results indicated that the best accuracy with the proposed method using the multilayer perceptron classifier was 98.74%, but without HIS, the accuracy for the same classifier was 92.23%.

By combining Word2vec with TFIDF, the accuracy of text classification could be increased, but the medical history of patients was important in the diagnosis of disease and could improve the accuracy. The results indicated that one should not focus only on medical reports and other clinical information and patients' history should also be used. Therefore, the use of HIS along with medical text reports could improve BI-RADS classification and have a positive effect on diagnosis and treatment processes.

Extracting information from the heart sound signal and detecting the abnormal signal in the early stage can play a vital role in reducing the death rate caused by cardiovascular diseases. Therefore, many researches have been done in processing these signals up to now. So, this study aimed to determine the improvement of heart abnormalities diagnosis by extracting features from the heart sound signal by applying machine learning classification algorithms.

The present descriptive–analytical study was conducted at Kazerun Azad University in 2023. The research data were selected from the 2016 Physionet Challenge database. After pre-processing and noise removal, 6 new features and 35 features (41 features) used in previous researches were extracted from the heart sound signals. The 6 new features are " Relative Average Perturbation", " five-point Period Perturbation Quotient", "local shimmer (in dB)", " three-point Amplitude Perturbation Quotient " and " five-point Amplitude Perturbation Quotient " and " correlation of time center of signal and frequency center of signal". The extracted features were applied as input to four classifiers of random forest, support vector machine, K nearest neighbor and linear discriminant analysis. Accuracy, sensitivity and specificity of each classification were calculated. In order to investigate the impact of new features in the diagnosis of cardiac abnormalities, the results obtained were compared with studies that used similar data and classifications but extracted fewer features from the data. The collected data were analyzed using t-tests and logistic regression.

The highest accuracy and sensitivity were obtained in the Linear Discriminant Analysis classifier, which are 91.52 and 96.19, respectively. The highest specificity was obtained in the Random Forest classifier at the rate of 88.90. According to the obtained results, by adding new features, the three indices of accuracy, sensitivity and specificity are improved in the two classifiers of K-nearest neighbor and Linear Discriminant Analysis. Extraction of these features also increases the level of specificity in the Random Forest classification.

The results indicated that the extraction of new features led to increase in the accuracy, sensitivity and specificity in the diagnosis of cardiac abnormalities compared to the results of previous researches.

Autism spectrum disorder is one of the psychological disorders in children. Timely and accurate diagnosis of this disorder is of great importance in providing appropriate care and treatment for children. The main objective of this research is to emphasize the significance of features related to autism and their diagnosis using an intelligent model, as some of these features have higher priority.

For this purpose, the principal component analysis (PCA) method was employed to prioritize the features, and after extracting the optimal features, automatic disease diagnosis was performed using artificial neural networks.

The data used in this study were collected from the Kaggle dataset, including 1054 individuals, out of which 728 were diagnosed with autism and 326 were healthy. The results of this study indicate that the gradual elimination of features and reduction from 18 to 12 features can lead to achieving the same accuracy in diagnosing the autism spectrum using artificial neural networks.

Reducing the number of features in artificial intelligence models for autism diagnosis not only improves and optimizes the diagnostic process but also helps in reducing parental stress and preserving their privacy due to the reduced number of questions. Ultimately, this leads to the generation of models with better performance and interpretability.

COVID-19 is a new virus that causes infection in the upper respiratory tract and lungs, and the number of deaths due to the disease has increased daily on the scale of a global epidemic. Chest X-ray images have been useful for monitoring various lung diseases and have recently been used to monitor COVID-19 disease.

In this research, a multi-stage process was used to recognize COVID-19 from X-ray images. In the first stage, pre-processing was done to normalize the data. In the second step, which is the most important step of the proposed method, feature extraction was done. The feature extraction operation was based on deep learning networks. After feature extraction, machine learning algorithms were used to classify images. The algorithms used in this section are support vector machine, nearest neighbor, and decision tree algorithms. The results of these categories are combined in the fourth step based on the majority vote.

The parameters used in this research are among the classification parameters, including precision, accuracy, recall, and F-criterion, which were obtained as 96.5, 92.25, 94, and 93, respectively.

The results of the experiments show the acceptable efficiency of the proposed method because, in addition to reducing the calculations by the separable layer, the combination of categories and their weighting has been used to obtain the final result.

Thermography is a non-invasive imaging technique that can be used to diagnose breast cancer. In this study, a method was presented for the extraction of suitable features in dynamic thermographic images of breast. The extracted features can help classify thermographic images as cancerous or healthy.

In this descriptive-analytical study, the images were taken from the IC/UFF database. A total of 196 people, including 41 cancer patients and 155 healthy individuals were investigated. Each person had 10 thermographic images and in total, 1960 images were analyzed. The images were captured using the FLIR ThermaCam S45 camera. The proposed model was presented based on a series of breast thermographic images of an individual to extract 8 suitable features.  The extracted features included mean, standard deviation, entropy, kurtosis, homogeneity, energy, skewness, and variance.

The extracted features were evaluated by the classifiers including the decision tree, support vector machine, quadratic symmetric analysis, and K-nearest neighbor algorithm using the ten-fold cross validation. The accuracy and sensitivity were 99% and 99.33% for decision tree algorithm, 98.46% and 95.12% for support vector machine algorithm, 100% and 100%, and 99% and 97.56% for K-nearest neighbor algorithm.

The results of this study showed that among the first-order statistical features, mean difference, skewness, entropy, and standard deviation are the most effective features which help to detect asymmetry. The features extracted by the proposed model can help classify the individuals into healthy or cancer-affected by thermal images.

Emotions play an important role in health, communication, and interaction between humans. The ability to recognize the emotional status of people is an important indicator of health and natural relationships. In DEAP database, electroencephalogram (EEG) signals as well as environmental physiological signals related to 32 volunteers are registered. The participants in each video were rated in terms of level of arousal, capacity, liking/disliking, proficiency, and familiarity with the video they watched.

In this study, a practical empirical method was adopted to classify capacity, arousal, proficiency, and interest by ranking the features extracted from signals using algorithms on EEG signals and environmental physiological signals (such as electromyography (EMG), electrooculography (EOG), galvanic skin response (GSR), respiration rate, photoplethysmography (PPG), and skin temperature. After initializing the signals from the database and pre-processing them, various features in the time and frequency domain were extracted from all signals. In this study, SVM and KNN classifiers, K-means clustering algorithm, and neural networks, such as PNN and GRNN were used to identify and classify emotions.

It was indicated in this study that the results of the classification of emotions using various methods and classifiers were well-established with high accuracy. The best accuracy results were obtained by applying the proposed method using SVM classifier based on features extracted from environmental signals (85.5%) and EEG signals (82.4%).

According to the results of the classification of emotions in this study, the proposed algorithm provides relatively better results compared with previous similar methods.

Breast cancer is the most common type of cancer and the second leading cause of cancer death among American women. In Iran, the rate of breast cancer is lower than in industrialized and western countries, but with the growing trend, it is predicted that breast cancer will become one of the most common cancers in the country in the future. Mammography is currently one of the most effective and popular methods for screening and diagnosing breast cancer. Breast cancer is the primary and most common disease found in women which causes second highest rate of death after lung cancer. The digital mammogram is the X-ray of breast captured for the analysis, interpretation, and diagnosis. Automatic detection of breast cancer in mammograms is a challenging task in Computer Aided Diagnosis (CAD) techniques. This article aims to provide an automated computer diagnostic system to help diagnose early breast cancer. First, breast cancer and the survival statistics of patients with it, breast imaging techniques, and the presence of symptoms that are present in the images are signs of the disease. In the following, by introducing important and efficient methods in designing automatic diagnostic systems and its structure in order to distinguish cancerous images from non-cancerous breasts, the results obtained from this research and validations have been presented.

Breast cancer, one of the most common cancers in women, has a high mortality rate. Providing a medical assistance system for early detection of abnormalities associated with this cancer will greatly assist pathologists in identifying the causes of the disease and increase performance and accuracy in diagnosis. Studying the background of research in this field to better understand the problem and how to design this system in different ways gives us a more accurate view of this issue and also defines the design challenges of such a system. The results obtained by the methods presented in this paper are abbreviated as BMD_ML. A total of 64 mammograms, 23 cancer images containing benign and malignant masses, and 41 non-cancerous images were used to evaluate the methods used in this paper, and the results were obtained from the inputs of these images. A combination of digital image processing methods, random statistics and machine learning methods is used to perform the pre-processing, segmentation and extraction of ROI, feature extraction and classification at the lowest error rate. CAD is used in mammography screening. Mammography screening is used to detect early breast cancer. The CAD system helps diagnose lesions and classify benign and malignant tumors. This system is mostly used in the United States and the Netherlands. The first CAD system for mammography was developed during a research project at the University of Chicago. CAD systems, despite their high sensitivity, have very few features; This makes the benefits of using CAD unclear. In this report we present a methodology for breast cancer detection in digital mammograms. Proposed methodology consists of three major steps like segmentation of breast region, removal of pectoral muscle and classification of breast muscle into cancerous and normal tissues.

This article aims to provide an automated computer diagnostic system to help diagnose early breast cancer. First, breast cancer and the survival statistics of patients with it, breast imaging techniques, and the presence of symptoms that are present in the images are signs of the disease. Then, important and efficient methods in designing automatic diagnostic systems and its structure are introduced and the work done in the past is examined by researchers active in this field. Finally, the techniques used in this paper are presented in order to distinguish between cancerous and non-cancerous breast cancer images. Segmentation of breast muscle was performed by employing Otsus segmentation technique, afterwards removal of pectoral muscle is carried out by seed selection and region growing technique. In next step, Gray Level Co-occurrence Matrices (GLCM) was created form which several features were extracted. At the end, several classifiers were trained to classify breast region into normal and cancerous tissues. The proposed classifier reports classification accuracy of 100 % for ANN and 96.3 % for decision tree algorithms (C5.0 and CHAID). Proposed methodology was validated on Mini-MIAS database and results were compared with previously proposed techniques, which shows that proposed technique can be reliably apply for breast cancer detection. Classification includes the final stage of designing such a system. Machine learning techniques have good performance for classifying tissue features obtained from mammograms. Machine learning is generally divided into two categories, supervised and uncontrolled. Learning without supervision requires a large amount of data to train the network. The classification methods used in this dissertation are one of the supervised methods, so that when creating a feature vector matrix, a column is assigned to whether the data is cancerous or non-cancerous. This method both speeds up learning and compares classified data with the predetermined target value when testing them. In this paper, several machine learning methods will be used to classify methods for classifying and comparing diagnostic accuracy. In this paper, classifications are performed in SPSS modeler software.

Numerous methods for extracting features were provided in the Overview of Features Extraction section. The solution presented in this article is to use a GLCM matrix. The matrix of gray surfaces always gives rise to different combinations of the brightness of the pixels in the image. CAD systems, despite their high sensitivity, have very few features; This makes the benefits of using CAD unclear. It can concluded that CAD could not have a significant effect on cancer detection rates, but it would inadvertently increase the recall rate (ie, the FP rate); However, various studies have shown significant inconsistencies in recall effects. In the design of the CAD system, the partitioning and extraction of the feature are of special importance. It should be noted to what extent the extraction characteristics describe the segmented area. In the future, in order to increase the accuracy of classifications and commercialization of this system, flexible features can be extracted from the image, which in addition to the lack of overlap between the features can be very compatible with machine learning methods. It is also possible to classify between types of anomalies, and after diagnosing whether the image is cancerous or non-cancerous, the type of anomaly associated with it can be identified.

Abstract In recent years, interest in research into the application of intelligent algorithms for diagnosis and categorization of diseases, especially cancer has increased dramatically. Tumor classification is an important task in medical diagnosis. Technological calculations are important due to their classification function in diagnosis of medical illnesses. Diagnosing and classifying medical images is a challenging task. To detect the malignancy of prostate cancer and the opioid or malignant breast cancer, deep neural network classifier, which is based on Tensor flow framework and Keras library, is used. In the training phase, educational images are considered along with the output class for the network. During training, the weight of the filter is updated every time. However, after several replications, optimal weights are updated and the network is trained to extract the best feature from the images. In this research, the proposed method due to using deep neural network and accurate feature extraction provides detection accuracy about 95.83% and 99.5% for breast and prostate cancers, respectively, which is more than 7% compared to other methods. Cancer is one of the most prevalent diseases in the world. Cancer is started from the cells, which are the basic building blocks making the tissue. One of the challenges in medical diagnostic techniques is the difficulty in analyzing dense tissues. Since the detection of the diagnosis by human is time-consuming and has a higher probability of error, the researchers have been trying to detect it automatically by using different algorithms.

By increasing life expectancy in the world, especially in developed countries, Alzheimer’s disease has become one of the most important and costly diseases. It is estimated that the prevalence of dementia will double every 5 years for people over 60 years. Unfortunately, this disease and other types of dementias impose a heavy burden on the economies of societies. In this study an automated method is presented and evaluated for Alzheimer disease detection in order to assist physician diagnosis. Methods and Materials: Magnetic resonance images of 236 patients were used. First, the gray and white matter regions and cerebrospinal fluid were extracted by image processing methods and their area were calculated and considered as feature vector for classification. Then, the support vector machine and neural networks were used to classify the Alzheimer patients from healthy ones. The gray matter region areas were significantly lower in Alzheimer group than healthy one. Using the regions area as feature vector and by means of support vector machine and neural networks as classifiers, a detection accuracy of 85% was achieved. Discussion and Alzheimer disease reduces the area of gray matter in brain which this changes are irreversible. Intellectual methods can detect this changes and diagnose Alzheimer disease. This methods could be used to assist physician for detection of disease.