PARENT SESSION
1:30 PM to 3:30 PM
Saturday, April 20, 2002
Poster Session 1 Noninvasive Treatment Monitoring and Treatment Planning
Room: Nevada 1-2
(MP01-9) Breast Perfusion Reconstruction Using Fractal Interpolation Functions.
Craciunescu, Oana*,1, Das, Shiva1, Vujaskovic, Zeljko1, Wong, Terry2, Samulski, Thaddeus1, 1 Duke University Medical Center, Durham, NC2 Duke University Medical Center, Durham, NC
ABSTRACT-
Thermal modeling for hyperthermia breast patients can provide relevant information to better understand the temperatures achieved during treatment. However, human breast is very perfused, making knowledge of the perfusion crucial to the accuracy of the temperature computations. It has been shown that the perfusion of blood in tumor tissue can be approximated using the relative perfusion index (RPI) determined from dynamic contrast-enhanced magnetic resonance imaging (DE-MRI). It was also concluded that the 3D reconstruction of tumor perfusion can be performed using fractal interpolation functions (FIF). The technique used was called piecewise hidden variable fractal interpolation (PHVFI). Changes in the protocol parameters for the dynamic MRI sequences in breast patients allowed us to be able to acquire more spatial slices, hence the possibility to actually verify the accuracy of the fractal interpolation. The MR protocol consisted of a series of enhanced Fast Gradient Echo 3D images. The series were initialized prior to injection of 2 ml/s Gd-PD (Magnevist), and continued for 30 minutes after. Sixty-eight slices of 4 mm thickness and 512x512 resolution were obtained every 63 s. This set was further stripped down to 34 slices. For both sets, the image intensity was analyzed on a pixel-by-pixel basis, and RPI values were determined from a double exponential fit of time vs. signal intensity. The RPI maps in the 34-slices set were interpolated using the PHVFI method. The interpolated slices were compared to the imaged slices in the original 68-slices set. The relative difference between the reconstruction and the original slice varied from 2 to 5%. Significantly, the fractal dimension of the interpolated slices is within 1-2% from the original images, thus preserving the fractality of the perfusion maps. The use of such a method becomes crucial when tumor size and imaging restrictions limits the number of spatial slices, requiring interpolation to fill the data between the slices.
KEYWORDS: brest tumors, relative perfusion index, magnetic resonance imaging, fractal interpolation functions