Transform domain algorithms for biomedical signal and image processing problems
Başlık çevirisi mevcut değil.
- Tez No: 400548
- Danışmanlar: DR. KAREN EGIAZARIAN
- Tez Türü: Doktora
- Konular: Acil Tıp, Göğüs Hastalıkları, Kardiyoloji, Emergency Medicine, Chest Diseases, Cardiology
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 2003
- Dil: İngilizce
- Üniversite: Tampereen Teknillinen Yliopisto (Tampere University of Technology)
- Enstitü: Yurtdışı Enstitü
- Ana Bilim Dalı: Belirtilmemiş.
- Bilim Dalı: Belirtilmemiş.
- Sayfa Sayısı: 158
Özet
Özet yok.
Özet (Çeviri)
In this thesis, the following problems are addressed: efficient de-noising of signals and images, image enhancement, de-noising of particular Electrocardiography (ECG) and Magnetocardiography (MCG) signals, simultaneous de-noising and enhancement of diagnostic X-ray images. Signal and image de-noising, which aims to find an estimate of the original signal from the corrupted data, is an important problem of signal processing. There exist various transform domain algorithms based on performing an orthogonal transformation, modifying transform coefficients and inverse transforming the modified transform coefficients. Those filters approach an optimal filter, whose parameters are not known in advance, under some circumstances. In this thesis, existing transform domain signal and image de-noising algorithms are equipped with adaptive bandwidth and base selection algorithms. Hence, they are made more adaptive to the local variations of the signals and images, thus improving the overall de-noising efficiency. The results are submitted in the included publications. ECG and MCG are fully noninvasive, totally harmless, safe and quick methods for measuring the electrical activity of the heart. However, ECG and MCG signals, especially MCG signals are corrupted by various noise sources. There are various methods for attenuating noise components from ECG and MCG signals. In this thesis transform domain algorithms are developed for removing the Electromyography (EMG) interference from ECG signals aimed at detail preservation and removing the residual random noise components from MCG signals. The most important advantage in the de-noising of particular signals is the limitedness of possible signal waveforms. Hence, general statistical properties of the signals can be incorporated into the de-noising algorithm. For ECG de-noising aimed at detail preservation the interband dependencies of the signals are utilized. For the final de-noising of the MCG signals, band specific thresholding is employed instead of using a fixed threshold for modification of spectral coefficients. The results are submitted in the included publications. Image enhancement concerns transforming the image into a form which is more suitable for human interpretation and machine analysis. Even though there exist many different methods of image enhancement, a unifying theory of image enhancement does not exist. Hence, there is no unique approach to the image enhancement problem, the most suitable approach depends on a particular image enhancement problem. An important issue to consider in enhancement of diagnostic X-ray images is the need of providing an image which is preferred by human expert since the final diagnosis will be made by the clinician based on the processed X-ray images. An important problem here is translating the subjective requirements of human experts into the properties of mathematical operations applied to the values corresponding to the brightness levels of the pixels of the images. In this thesis a Wavelet transform (WT) based simultaneous diagnostic X-ray image de-noising and enhancement algorithm was developed, utilizing the approximation of WT detail coefficients to the local contrasts at each resolution. The quality control procedure for the enhanced images is explained in the included publications. A histogram modification algorithm, which does not distort the visual information and which can be cascaded with the sharpening algorithm is also developed. Since improving the relative significance of particular details and improving the usage of the visual span are two different approaches to the image enhancement problem, improving the image in both senses provide better final image quality.
Benzer Tezler
- Image reconstruction with deep learning and applications in MR images
Derin öğrenme ile görüntü geriçatımı ve MR görüntülerinde uygulamaları
AMIR AGHABIGLOU
Doktora
İngilizce
2022
Mekatronik Mühendisliğiİstanbul Teknik ÜniversitesiMekatronik Mühendisliği Ana Bilim Dalı
PROF. DR. ENDER METE EKŞİOĞLU
- Gevşeme temelli kenar belirleme algoritması
Başlık çevirisi yok
GÜRAY GÜNGÖR
Yüksek Lisans
Türkçe
1998
Elektrik ve Elektronik Mühendisliğiİstanbul Teknik ÜniversitesiBiyomedikal Mühendisliği Bilim Dalı
DOÇ. DR. TAMER ÖLMEZ
- Dalgacık dönüşümündeki yerel tepe değerlerinden görüntü geriçatma ve sıkıştırma
Image reconstruction from the wavelet maxima and coding
HÜSEYİN ERGÜN
Yüksek Lisans
Türkçe
1997
Elektrik ve Elektronik Mühendisliğiİstanbul Teknik ÜniversitesiElektronik ve Haberleşme Mühendisliği Ana Bilim Dalı
DOÇ. DR. MUHİTTİN GÖKMEN
- EEG sinyalleri ve derin öğrenme modelleri kullanılarak dikkat eksikliği ve hiperaktivite bozukluğu tespiti
Attention deficit hyperactivity disorder prediction using EEG signals and deep learning models
YAĞMUR CAN
Yüksek Lisans
Türkçe
2024
BiyomühendislikTOBB Ekonomi ve Teknoloji ÜniversitesiBiyomedikal Mühendisliği Ana Bilim Dalı
PROF. DR. OSMAN EROĞUL
- Biyolojik işaretlerin gelişmiş bir sayısal işaret işlemcisiyle işlenmesi
Biomedical signal processing using a high performance DSP
DERYA DEMİR
Yüksek Lisans
Türkçe
1991
Elektrik ve Elektronik Mühendisliğiİstanbul Teknik ÜniversitesiPROF.DR. ERTUĞRUL YAZGAN