國立虎尾科技大學 |

Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement./
Author:	Rasul, Raisa Binte.
Description:	1 online resource (132 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-07, Section: B.
Contained By:	Dissertations Abstracts International85-07B.
Subject:	Biomedical engineering. -
Online resource:	click for full text (PQDT)
ISBN:	9798381413908

Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement.
Rasul, Raisa Binte.

Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement. - 1 online resource (132 pages)

Source: Dissertations Abstracts International, Volume: 85-07, Section: B.

Thesis (Ph.D.)--Cornell University, 2023.

Includes bibliographical references

Standard image analysis is typically performed manually, which is often time-inefficient and in many cases imprecise. This is an issue particularly in the field of radiology where clinical decisions need to be made both quickly and accurately based on medical image analysis results. Therefore, automated image analysis has been a field of high interest over the past several decades. In this thesis, work is described in which different automated methods of image analysis are used to quantify automated image measurement precision, to segment the amount of calcium present in the coronary arteries, and to identify locations of asbestos burden along the pleural surface of the lung.In the first section a novel method of quantifying the dynamic measurement precision of an automated system measuring spherical, soot-producing fuel droplets undergoing combustion is presented. This automated measuring system utilizes least-squares circle/ellipse fitting to detect and measure the diameter of these droplets even when partially occluded by soot. From this, quality metrics such as gradient intensity and goodness of fit were identified, and synthetic image simulation was used to correlate these quality metrics with dynamic measurement precision. The final measurement system was able to achieve measurement precision of ± 0.2 pixels in cases of low occlusion. This work demonstrates that an automated image analysis system is capable of performing measurements with dynamic subpixel precision, far higher than what is possible with human annotations.In the second chapter, an automated system for detecting and quantifying coronary artery calcium (CAC) in cardiac computed tomography (CT) scans is described. This work utilized images from the Multi-Ethnic Study of Atherosclerosis (MESA) dataset along with radiologist-labeled pixel ground truths for each scan. This algorithm utilizes traditional computer vision techniques to achieve this goal. Within this, a heart segmentation was developed that enabled a general localization of the coronary arteries by mapping a probability map onto a heart surface mesh. This was used to identify candidate CAC plaques. Features were then extracted, and a random forest classifier was used to determine true and false CAC. This method achieved a Pearson correlation of 0.98, an average Dice score of 0.85, and a risk-category accuracy of 0.87. This work was able to achieve results similar to existing state-of-the-art methods.In the final chapter, an algorithm for detecting the presence of asbestos exposure in full-chest CT scans is presented. This work focused on patients that live in Libby, Montana where a significant percent of the population has been exposed to asbestos originating from nearby vermiculite mines. Asbestos leads to a multitude of presentations on the lung surface including the thickening of the pleural membrane and manifestation of plaques. Specifically, in Libby cases, patients exhibit a specific type of pleural thickening, like lamellar pleural thickening, on chest CT scans that is difficult to detect by radiologist. Therefore, to detect this thickening and other disease conditions, the lung was segmented, and its surface was sampled in an orthogonal fashion to obtain subpixel intensity profiles along the edge of the lung. Features from these profiles were then used to identify regions of interest where there was likely asbestos exposure. This method was evaluated on healthy and diseased Libby cases as well as an external dataset [Lung Image Database Consortium (LIDC)]. The fraction of the lung surface identified as diseased by the algorithm was significantly higher in the diseased cases when compared to the healthy and LIDC cases. This showed that the methods developed for identifying pleural thickening in Libby cases was successfully achieved.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798381413908Subjects--Topical Terms:

588770
Biomedical engineering.
Subjects--Index Terms:

3D imagesIndex Terms--Genre/Form:

554714
Electronic books.

Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement.
LDR:05277ntm a22004097 4500 001 1151875
005 20241118135851.5
006 m o d
007 cr mn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798381413908
035 $a (MiAaPQ)AAI30816031
035 $a AAI30816031
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Rasul, Raisa Binte. $3 1478714
245 1 0 $a Image Analysis by Projections in 3D Images With Applications in Medical Image Analysis and Precise Image Measurement.
264 0 $c 2023
300 $a 1 online resource (132 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-07, Section: B.
500 $a Advisor: Reeves, Anthony.
502 $a Thesis (Ph.D.)--Cornell University, 2023.
504 $a Includes bibliographical references
520 $a Standard image analysis is typically performed manually, which is often time-inefficient and in many cases imprecise. This is an issue particularly in the field of radiology where clinical decisions need to be made both quickly and accurately based on medical image analysis results. Therefore, automated image analysis has been a field of high interest over the past several decades. In this thesis, work is described in which different automated methods of image analysis are used to quantify automated image measurement precision, to segment the amount of calcium present in the coronary arteries, and to identify locations of asbestos burden along the pleural surface of the lung.In the first section a novel method of quantifying the dynamic measurement precision of an automated system measuring spherical, soot-producing fuel droplets undergoing combustion is presented. This automated measuring system utilizes least-squares circle/ellipse fitting to detect and measure the diameter of these droplets even when partially occluded by soot. From this, quality metrics such as gradient intensity and goodness of fit were identified, and synthetic image simulation was used to correlate these quality metrics with dynamic measurement precision. The final measurement system was able to achieve measurement precision of ± 0.2 pixels in cases of low occlusion. This work demonstrates that an automated image analysis system is capable of performing measurements with dynamic subpixel precision, far higher than what is possible with human annotations.In the second chapter, an automated system for detecting and quantifying coronary artery calcium (CAC) in cardiac computed tomography (CT) scans is described. This work utilized images from the Multi-Ethnic Study of Atherosclerosis (MESA) dataset along with radiologist-labeled pixel ground truths for each scan. This algorithm utilizes traditional computer vision techniques to achieve this goal. Within this, a heart segmentation was developed that enabled a general localization of the coronary arteries by mapping a probability map onto a heart surface mesh. This was used to identify candidate CAC plaques. Features were then extracted, and a random forest classifier was used to determine true and false CAC. This method achieved a Pearson correlation of 0.98, an average Dice score of 0.85, and a risk-category accuracy of 0.87. This work was able to achieve results similar to existing state-of-the-art methods.In the final chapter, an algorithm for detecting the presence of asbestos exposure in full-chest CT scans is presented. This work focused on patients that live in Libby, Montana where a significant percent of the population has been exposed to asbestos originating from nearby vermiculite mines. Asbestos leads to a multitude of presentations on the lung surface including the thickening of the pleural membrane and manifestation of plaques. Specifically, in Libby cases, patients exhibit a specific type of pleural thickening, like lamellar pleural thickening, on chest CT scans that is difficult to detect by radiologist. Therefore, to detect this thickening and other disease conditions, the lung was segmented, and its surface was sampled in an orthogonal fashion to obtain subpixel intensity profiles along the edge of the lung. Features from these profiles were then used to identify regions of interest where there was likely asbestos exposure. This method was evaluated on healthy and diseased Libby cases as well as an external dataset [Lung Image Database Consortium (LIDC)]. The fraction of the lung surface identified as diseased by the algorithm was significantly higher in the diseased cases when compared to the healthy and LIDC cases. This showed that the methods developed for identifying pleural thickening in Libby cases was successfully achieved.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Biomedical engineering. $3 588770
650 4 $a Medical imaging. $3 1180167
650 4 $a Bioengineering. $3 598252
653 $a 3D images
653 $a Image measurement
653 $a Medical image segmentation
653 $a Polygon surface mesh
653 $a CT scans
653 $a Radiology
655 7 $a Electronic books. $2 local $3 554714
690 $a 0541
690 $a 0202
690 $a 0574
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Cornell University. $b Biomedical Engineering. $3 1193194
773 0 $t Dissertations Abstracts International $g 85-07B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30816031 $z click for full text (PQDT)