Institutional Repository of Vilnius University
Neuroniniai tinklai ir jų taikymas biomoksluose /
Abstract
dc:descriptionThe aim of this bachelor thesis is to analyze recursive neural networks and convolutional neural networks and their efficiency and accuracy to investigate how network parameters affect the accuracy of medical image recognition. To achieve this goal, the following goals have been set: 1) Review the literature related to RNN and CNN and their application in life sciences. 2) To deepen the knowledge about RNN and CNN and the principle of their operation. 3) To find and prepare the data sets required for the research. 4) Establish RNN and CNN networks. 5) Train RNN and CNN networks with prepared data. 6) Summarize the results and present conclusions. The main goal of the bachelor's thesis to deepen the knowledge about recurrent neural networks and convolutional neural networks and their use in the life sciences for the classification of medicinal images, as well as the creation of a recursive neural network and a convolutional neural network has been achieved. When In order to achieve the goal, it was necessary to discuss the theoretical part about machine learning, neurons, artificial neurons, artificial neural networks and what they consist of, discuss and delve into RNN and CNN and their operation, to find data for the investigation, to build networks and to carry out the investigation. Using the Python programming language and the Google colab programming environment I created neural networks that are capable of training with given data. Data that was used in the work was medical images, MRI images of the human brain that are designed to detect dementia. The study was performed with 4 neural networks and 3 data sets. In total there were 12 studies done. After describing the theoretical part and conducting the research, it can be stated that the research was successful. Networks successfully trained with the given data and gave quite good results. It is safe to say that CNN is more suitable for classifying images and can be used to classify medical images, to make diagnoses. It can be seen that dementia and its progression can be determined by neural networks. These studies are extremely important and provide a glimpse into the future of medicine, where diagnoses will become more and more accurate and it will be easier to make diagnoses.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Trijonytė, Justė,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD146240711&prefLang=en_US