Simulating metal ct artefacts for ground truth generation in deep learning.
(2023) BMEM01 20231Department of Biomedical Engineering
- Abstract
- CT scanning stands as one of the most employed imaging techniques used in clinical field. In the presence of metal implants in the field of view (FOV), distortions and noise appear on the 3D image leading to inaccurate bone segmentation, often required for surgery planning or implant design. In this research project, we focused on developing a pipeline to create a rich ground truth dataset, forming the foundation for training a deep-learning bone segmentation algorithm capable of great results even in the presence of metal artefacts.
In order to build an extensive artefact-contaminated database, metal-free datasets were collected and bone delineated using Segment 3DPrint automatic bone-segmentation tool. Realistic metal implants were... (More) - CT scanning stands as one of the most employed imaging techniques used in clinical field. In the presence of metal implants in the field of view (FOV), distortions and noise appear on the 3D image leading to inaccurate bone segmentation, often required for surgery planning or implant design. In this research project, we focused on developing a pipeline to create a rich ground truth dataset, forming the foundation for training a deep-learning bone segmentation algorithm capable of great results even in the presence of metal artefacts.
In order to build an extensive artefact-contaminated database, metal-free datasets were collected and bone delineated using Segment 3DPrint automatic bone-segmentation tool. Realistic metal implants were then designed in proper anatomical locations and synthetic artefacts generated using a MATLAB-based algorithm. The effectiveness of the simulator has been tested on real data using an anthropomorphic hand phantom with metal implants inserted and scanned with standard clinical CT parameters.
The simulator has proven to successfully mimic physical phenomena such as beam hardening, phantom starvation and noise which are the underlying causes of real metal artefacts. It produces realistic artefacts shapes, even for complex metal configurations. Additional datasets already exhibiting metal artefacts were also added to the database. The simulator was used there only to virtually rescanned those datasets for augmentation reasons.
Finally, a training pipeline was imagined using the artefact simulator in parallel to the training process. Data can thus be constantly augmented with new features as the training of the network is running. (Less) - Popular Abstract
- Heading: Deep-learning-driven augmentation of a computed tomography database with synthetic metal artifacts.
Introduction: A CT database was enriched with synthetic metal artefacts using custom 3D implants of various designs and attenuation properties. The aim is to improve the robustness of deep-learning algorithms.
Main text: CT-scanning is a major imaging technique widely utilized in the clinical field on a daily basis. This technique provides valuable information, including the structure and shape of bones, which is crucial for advanced diagnostics and surgical planning. The process of extracting tissue structures from CT scans is known as segmentation and relies on efficient technologies like deep-learning algorithms to achieve... (More) - Heading: Deep-learning-driven augmentation of a computed tomography database with synthetic metal artifacts.
Introduction: A CT database was enriched with synthetic metal artefacts using custom 3D implants of various designs and attenuation properties. The aim is to improve the robustness of deep-learning algorithms.
Main text: CT-scanning is a major imaging technique widely utilized in the clinical field on a daily basis. This technique provides valuable information, including the structure and shape of bones, which is crucial for advanced diagnostics and surgical planning. The process of extracting tissue structures from CT scans is known as segmentation and relies on efficient technologies like deep-learning algorithms to achieve rapid and precise segmentation results. However, challenges arise when metal objects, such as implants, are present within the CT-scanned area.
These metal objects cause significant image distortions and noise, commonly referred to as artefacts. These artefacts substantially degrade image quality and complicate the segmentation process due to the presence of intense dark and bright streaks that overlay the tissue structures.
Consequently, this thesis was dedicated to developing a pipeline that focuses on robust artefact-handling algorithms based on deep-learning networks. CT scans were collected, and synthetic artefacts were simulated using MATLAB programming environment. The artefact simulation process provides a versatile framework with adjustable parameters for generating artefacts of various shapes and intensities. This simulation framework serves as a valuable tool for potentially training a deep-learning network.
The simulator exhibited remarkable capability in generating realistic artefacts, encompassing complex shapes and intensities. Consequently, constitutes a rich input for the deep-learning algorithm to accurately segment bones from clinical cases that feature real artefacts. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9127413
- author
- Barakat, Arthur LU
- supervisor
- organization
- course
- BMEM01 20231
- year
- 2023
- type
- H2 - Master's Degree (Two Years)
- subject
- language
- English
- additional info
- 2023-07
- id
- 9127413
- date added to LUP
- 2023-06-27 11:39:36
- date last changed
- 2023-06-27 11:39:36
@misc{9127413,
abstract = {{CT scanning stands as one of the most employed imaging techniques used in clinical field. In the presence of metal implants in the field of view (FOV), distortions and noise appear on the 3D image leading to inaccurate bone segmentation, often required for surgery planning or implant design. In this research project, we focused on developing a pipeline to create a rich ground truth dataset, forming the foundation for training a deep-learning bone segmentation algorithm capable of great results even in the presence of metal artefacts.
In order to build an extensive artefact-contaminated database, metal-free datasets were collected and bone delineated using Segment 3DPrint automatic bone-segmentation tool. Realistic metal implants were then designed in proper anatomical locations and synthetic artefacts generated using a MATLAB-based algorithm. The effectiveness of the simulator has been tested on real data using an anthropomorphic hand phantom with metal implants inserted and scanned with standard clinical CT parameters.
The simulator has proven to successfully mimic physical phenomena such as beam hardening, phantom starvation and noise which are the underlying causes of real metal artefacts. It produces realistic artefacts shapes, even for complex metal configurations. Additional datasets already exhibiting metal artefacts were also added to the database. The simulator was used there only to virtually rescanned those datasets for augmentation reasons.
Finally, a training pipeline was imagined using the artefact simulator in parallel to the training process. Data can thus be constantly augmented with new features as the training of the network is running.}},
author = {{Barakat, Arthur}},
language = {{eng}},
note = {{Student Paper}},
title = {{Simulating metal ct artefacts for ground truth generation in deep learning.}},
year = {{2023}},
}