Fast iterative segmentation of high resolution medical images

Conference · Tue Dec 31 04:00:00 EST 1996

OSTI ID:513296

Hebert, T J ^[1]

---

1. Univ. of Houston, TX (United States)

Various applications in positron emission tomography (PET), single photon emission computed tomography (SPECT) and magnetic resonance imaging (MRI) require segmentation of 20 to 60 high resolution images of size 256x256 pixels in 3-9 seconds per image. This places particular constraints on the design of image segmentation algorithms. This paper examines the trade-offs in segmenting images based on fitting a density function to the pixel intensities using curve-fitting versus the maximum likelihood method. A quantized data representation is proposed and the EM algorithm for fitting a finite mixture density function to the quantized representation for an image is derived. A Monte Carlo evaluation of mean estimation error and classification error showed that the resulting quantized EM algorithm dramatically reduces the required computation time without loss of accuracy.

• OSTI ID: 513296
• Report Number(s): CONF-961123--
• Country of Publication: United States
• Language: English

Related Subjects

44 INSTRUMENTATION
INCLUDING NUCLEAR AND PARTICLE DETECTORS
55 BIOLOGY AND MEDICINE
BASIC STUDIES
99 GENERAL AND MISCELLANEOUS
ALGORITHMS
IMAGE PROCESSING
ITERATIVE METHODS
MAXIMUM-LIKELIHOOD FIT
MONTE CARLO METHOD
NMR IMAGING
TOMOGRAPHY