Naslov
4D CT Cardiac image enhancement for Subjective Medical Quality Perception of the Left Ventricle

Autori
Leventić, Hrvoje ; Livada, Časlav ; Galić, Irena ; Zlokolica, Vladimir ; Velicki, Lazar ; Babin, Danilo ; Obradović, Ratko ; Mihajlović, Bogoljub

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
MIPS2015 Book of Abstracts / - , 2015, 60-60

Skup
Medical Image Perception Society (MIPS) Conference XVI

Mjesto i datum
Gent, Belgija, 03.06.2015. - 05.06.2015

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
4D CT ; Image Enhancement ; Subjective medical quality perception ;

Sažetak
Rationale Recently it has been described that even though MRI heart imaging is the gold standard for cardiac structure and function evaluation, MRI is not always feasible in some patients. For such patients CT is a viable alternative, which however can introduce significant quality degradation, especially in preferable case of low level of radiation. In this paper we investigate and assess image enhancement approaches to be applied to image slices of 4D CT technology to facilitate decisions made by cardiologists. This applies to both medical doctor’s subjective visual quality opinion as well as semi-automatic computer vision tasks (in this case segmentation of heart left ventricle) which efficiency is again to be assessed by the medical doctors. The improved quality assessment by cardiologists should aid more precise detection of heart failure, for which left ventricle dysfunction is often the main cause. Methods We have processed 4D CT slices with image enhancement methods related to contrast enhancement and de-noising algorithms. The CT data bases were selected by medical doctors (MD), specialized in cardiac surgery. We explore the usage of existing multi-scale methodologies (Laplacian Pyramid and Non- decimated wavelet transform) for local image enhancement and de-noising but also develop our own algorithm for contrast enhancement and de-noising algorithm for this special case of cardiac images. The explored algorithms for image enhancement were applied with different parameters tunable my MDs, which have provided subjective quality assessment for the purpose of LV segmentation and 3D volume determination. Results Preliminary results indicate that our proposed MD-quality-assessment driven image enhancement algorithm significantly improves the image quality and its usability for more precise LV registration. Conclusions Medical quality assessment is highly important for image enhancement algorithm optimization.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA