University of Twente Student Theses

Login
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.

Convolutional autoencoders to process 4D gynaecological data

Hofsteenge, M. T. (2020) Convolutional autoencoders to process 4D gynaecological data.

[img] PDF
7MB
Abstract:Unsupervised deep learning is a great way to gain more understanding on data sets. We applied unsupervised methods to gain insight on 4D ultrasound data of the pelvic floor. We reduced the dimensionality of the 3 physical dimensions of the ultrasound with a convolutional autoencoder, in an unsupervised manner. This reduced the data from 4D to 2D, maintaining the time dimension. Every ultrasound shows a patient performing a maneuver, which is either contraction or valsalva. These are thought to be prevalent features in the ultrasounds. Using the dimensionality reduced data, we successfully classified the maneuver performed in the ultrasound, with supervised and unsupervised methods. The supervised classification resulted in 80-95% accuracy, and unsupervised in 75-90% accuracy. This demonstrates that useful data representations can be found in very large data by using an unsupervised convolutional autoencoder for dimensionality reduction.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:50 technical science in general
Programme:Electrical Engineering MSc (60353)
Link to this item:https://purl.utwente.nl/essays/82868
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page