List of articles (by subject)

---

• Open Access Article
  • Abstract Page
  • Full-Text
  
  1 - An optimal liver segmentation method in MRI images using adaptive water flow model
  Marjan Heidari Mehdi Taghizadeh Hassan Masoumi مرتضی ولی زاده
  Liver segmentation in medical images is still considered as a challenge in computer diagnosis systems. In this paper, an optimal algorithm based on the adaptive water flow model for segmentation is introduced. This algorithm first processes the image by means of a trans More

  Liver segmentation in medical images is still considered as a challenge in computer diagnosis systems. In this paper, an optimal algorithm based on the adaptive water flow model for segmentation is introduced. This algorithm first processes the image by means of a transfer function designed based on the probability distribution function of the brightness levels of the liver pixels to distinguish the liver region from the rest of the parts. Then, with the help of the rainfall algorithm, which is controlled based on the spatial information and light levels of the liver, possible areas of the liver are extracted, and further, the possible areas of the liver are classified with a layered perceptron neural network, using shape and texture features. Classification of areas instead of pixels has increased the efficiency of the algorithm. The obtained experimental results show a far more appropriate performance in comparison with other evaluation algorithms Manuscript profile
• Open Access Article
  • Abstract Page
  • Full-Text
  
  2 - Detection of healthy and unhealthy cardiac signals based on deep learning using convolutional neural network
  Alireza nasrabadian mohammad amin nooshzadeh madiha abbas zadeh barani mohammad mahdi moradi
  According to the statements of the World Organization, the most important factor threatening humans is cardiac arrhythmias. According to the latest global health statistics, nearly 50% of deaths are due to heart diseases. According to research, 25% of deaths due to hear More

  According to the statements of the World Organization, the most important factor threatening humans is cardiac arrhythmias. According to the latest global health statistics, nearly 50% of deaths are due to heart diseases. According to research, 25% of deaths due to heart diseases can be revived with timely and correct diagnosis. The electrocardiogram signal is the most important and dependent signal related to the heart. Registration of this signal is low-cost, fruitful, and powerful in detecting arrhythmias. Feature extraction is the most important part of recognition and processing. Deep features based on convolutional neural networks are very powerful and can be done without manual intervention. In this article, deep features are extracted using deep learning based on a convolutional neural network. Then the classification results were calculated with an average accuracy of 99.3% and an average sensitivity of 99.1% with 10-fold cross-validation. According to the obtained results, it can be said that the proposed method has the ability to classify cardiac arrhythmias with acceptable accuracy. Manuscript profile