東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Wavelet-based pulmonary nodules feat...

Osicka, Teresa.

FindBook

Google Book

Amazon

博客來

Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans./
作者:	Osicka, Teresa.
面頁冊數:	307 p.
附註:	Adviser: Farid Ahmed.
Contained By:	Dissertation Abstracts International68-12B.
標題:	Biophysics, Medical. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3294909
ISBN:	9780549395980

Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans.
Osicka, Teresa.

Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans. - 307 p.

Adviser: Farid Ahmed.

Thesis (Ph.D.)--The Catholic University of America, 2008.

One of the fundamental issues in medical image processing is the texture characterization of intrinsic structures; specifically feature generation, extraction and classification with regard to discrimination between normal and abnormal tissue characteristics. This thesis is concerned with the analysis and classification of benign and malignant lung nodules extracted from computed tomography (CT) scans, where the pixel intensity and structure of the nodule texture are not clearly determinable, hence signal processing and statistical approaches are appropriate. We investigate the advantages of multi-scale/multi-resolution signal processing methods namely discrete wavelet transform (DWT), higher order statistical methods, computationally low complexity segmentation method and the visualization tool called "Heat Maps". Consequently these advantages are combined to form novel approach to lung nodules tissue characterization.

ISBN: 9780549395980Subjects--Topical Terms:

1017681
Biophysics, Medical.

Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans.
LDR:03361nam 2200313 a 45 001 949341
005 20110525
008 110525s2008 ||||||||||||||||| ||eng d
020 $a 9780549395980
035 $a (UMI)AAI3294909
035 $a AAI3294909
040 $a UMI $c UMI
100 1 $a Osicka, Teresa. $3 1272722
245 1 0 $a Wavelet-based pulmonary nodules features characterization on computed tomography (CT) scans.
300 $a 307 p.
500 $a Adviser: Farid Ahmed.
500 $a Source: Dissertation Abstracts International, Volume: 68-12, Section: B, page: 8143.
502 $a Thesis (Ph.D.)--The Catholic University of America, 2008.
520 $a One of the fundamental issues in medical image processing is the texture characterization of intrinsic structures; specifically feature generation, extraction and classification with regard to discrimination between normal and abnormal tissue characteristics. This thesis is concerned with the analysis and classification of benign and malignant lung nodules extracted from computed tomography (CT) scans, where the pixel intensity and structure of the nodule texture are not clearly determinable, hence signal processing and statistical approaches are appropriate. We investigate the advantages of multi-scale/multi-resolution signal processing methods namely discrete wavelet transform (DWT), higher order statistical methods, computationally low complexity segmentation method and the visualization tool called "Heat Maps". Consequently these advantages are combined to form novel approach to lung nodules tissue characterization.
520 $a We developed practical aspects, based on the discrete wavelet transform (DWT) theoretical background, for the feature generation for characterization of benign and malignant lung nodules signal, where wavelets coefficients and decomposition properties are integrated into set of features to increase the discrimination power and, consequently classification performance. Furthermore, to overcome practical difficulties of traditional segmentation approaches that require separation of a very small object from a background, we introduced squared windows for the segmentation purposes within the novelty of the lung nodule characterization framework. This demands only fast algorithm to extract the nodule. The "Heat Maps", in turn, have the ability to identify and visualize a heterogeneity/homogeneity between nodules and nodule clusters thus, to provide useful information for radiologists for further investigation to understand why some of benign nodules look like malignant ones.
520 $a The proposed scheme is compared to higher order statistical methods commonly used for texture characterization, with a use of the same original image dataset from the National Lung Screening Trial (NLST) of the National Cancer Institute (NCI). The results are then analyzed in order to evaluate lung nodules classification performance based upon accuracy, computational complexity, robustness to noise, and discriminative power of selected features with the "Heat Maps" visualization tool.
590 $a School code: 0043.
650 4 $a Biophysics, Medical. $3 1017681
650 4 $a Computer Science. $3 626642
650 4 $a Health Sciences, Radiology. $3 1019076
690 $a 0574
690 $a 0760
690 $a 0984
710 2 $a The Catholic University of America. $3 1019220
773 0 $t Dissertation Abstracts International $g 68-12B.
790 $a 0043
790 1 0 $a Ahmed, Farid, $e advisor
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3294909