University of Twente Student Theses

Login

Automatic Segmentation and 3D Reconstruction of Liver and Tumor

Wardhana, Girindra (2018) Automatic Segmentation and 3D Reconstruction of Liver and Tumor.

This is the latest version of this item.

[img] PDF
3MB
Abstract:In liver-related symptoms, image segmentation is an important step to help clinicians in visualizing the anatomy of the patient’s. A deep convolutional neural network is proposed to perform automatic liver and tumor segmentation from CT images. The network structure utilizes the encoder and decoder structure from the SegNet with some modifications. Several tests have been conducted to examine the network, including the preparation on the dataset, the variation of network configuration and the evaluation using manual segmentation. The test result reveals that different preparation techniques affect the segmentation accuracy. At the same time, utilizing class balance in the network is very crucial, where the network without class balance ignores the tumor from the image. In the LiTS challenge, the proposed network outperforms other states of the art methods for detecting the tumor and subsequently get the first rank on the online LiTS leaderboard in Recall category. Meanwhile, the proposed network also shows an impressive performance, where the network segmentation exceeds the manual segmentation result in term of segmentation accuracy and processing time. Further work of this project should be focused on improving the segmentation performance, such as implementing liver detection and tumor detection.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:01 general works, 50 technical science in general, 54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:https://purl.utwente.nl/essays/76513
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page