Avtor/Urednik		Klemenčič, Jan; Valenčič, Vojko; Pečarič, Nuška
Naslov		Volumetrija hipokampusa na osnovi slik magnetne resonance
Tip		članek
Vir		In: Zajc B, editor. Zbornik 9. elektrotehniške in računalniške konference ERK 2000. Zvezek B. Računalništvo in informatika, umetna inteligenca, robotika, razpoznavanje vzorcev, biomedicinska tehnika, močnostna elektrotehnika, didaktika, študentski članki; 2000 sep 21-23; Portorož. Ljubljana: Slovenska sekcija IEEE,
Leto izdaje		2000
Obseg		str. 285-8
Jezik		slo
Abstrakt		Hippocampus is a temporal lobe brain structure. Its atrophic condition is related to some brain diseases, especially epilepsy. Usually, its volume and shape are visually evaluated from coronal T1 weighted MR cross-sections. In this paper, we present an active contour based semi-automatic hippocampus segmentation algorithm. The algorithm is intuitive to neuroradiologist, as it uses the same coronal MR cross-sections for initalization. Initalization consists of manual segmentation of some key cross-sections. We discus numeroussegmentation-related problems. Images have low contrasts and low S/N ratios. Hippocampus boundaries are surrounded by multiple other edges.Also, they are not always continuous, therefore the deformable active contours principle was chosen for segmentation. The active contours inherent limitations were overcome by encoding a priori shape information into the deformation algorithm. We present the results obtained by segmenting normal and atrophic hippocampi. We experimentally prove, that all the algorithm parameters may only be set once for a particular type of application. In comparison to manual segmentation, our method reduces the workload on the neuroradiologist by about four times.
Deskriptorji		HIPPOCAMPUS MAGNETIC RESONANCE IMAGING IMAGE PROCESSING, COMPUTER-ASSISTED