List of Articles آنالیز خودکار تصاویر

• Open Access Article
  • Abstract Page
  • Full-Text
  
  1 - A Novel Semi-Supervised Technique for Selecting Appropriate Sperm in Infertility Treatment
  Asefeh Tavakkoli Seyed Abolghasem Mirroshandel Fatemeh Ghasemian
  Introduction:Nowadays infertility is recognized as one of the most common clinical problems around the globe, and one of the most worrying social issues in different cultures and societies. In the meantime, efforts have always been made to prevent the progression of inf More

  Introduction:Nowadays infertility is recognized as one of the most common clinical problems around the globe, and one of the most worrying social issues in different cultures and societies. In the meantime, efforts have always been made to prevent the progression of infertility caused by this factor by carefully examining the most effective male factor - as one of the potential parties in infertility problems - that is, analyzing the quantity and quality of sperm. On one hand, traditional methods have lots of problems such as inadequate accuracy, clinicians' disagreements, and prolonged treatment. On the other hand, the successes of machine learning in many areas prompted researchers to move toward automating sperm morphology analysis by means of machine learning.Methods:The ladder network as a semi-supervised learning algorithm, by using a small number of labeled samples and a larger part of unlabeled data, shows suitability and compliance with the real-world requirements in this field of study. In this regard, in order to implement ladder networks, the structure of stack noise removal auto-encoders with the architecture of two parallel encoders has been used to represent the samples and a decoder to reconstruct the samples. The present study by applying changes and improving various factors, especially input noise, has obtained good results in the analysis of low-resolution images without coloring.Results:The proposed model succeeded by extracting positive and fruitful features from the images of the head, acrosome, and vacuole of human sperm, showing an acceptable accuracy for classifying them into two natural and abnormal classes, and finally selecting the appropriate sperm to participate in the artificial insemination process. The study of the proposed model for all three sperm sections (head, vacuole, and acrosome) succeeded, despite low-quality images, achieving impressive results of more than 70% for the head and acrosome and more than 80% for the vacuole.Conclusion: In the future, we intend to improve the proposed model by finding ways to increase the accuracy and reduce the error of the test results and show that the change in the type of noise or how it is applied to the network will have a significant impact on the network performance. Manuscript profile