利用頻率空間三度空間形變技術做腦部模型均質化

Abstract

腦部功能的研究在醫學領域上一直是重要的課題之一，腦功能與組織之對應的大腦地圖需要一個標準腦。本文試圖以物體形變的技術，建立一個平均腦。腦部資料取自於核磁共振造影（MRI）人腦影像及共軛焦顯微鏡（Confocal microscope）擷取的昆蟲腦影像，經過前置處理後擷取出腦部結構的體積模型，將兩組腦部結構經過校準（alignment），使兩組腦部體積模型能夠對齊在一起；接著把兩組模型轉換到頻率空間進行形變的處理，並透過適當的濾波器及參數來控制均質模型的效果。實驗結果顯示了校準對於均值化的正面影響，並且探討比較不同參數的效果。

Brain mapping is one of the important problems in biology and medicine. The brain mapping research needs a standard brain. We try to construct a standard brain from Magnetic Resonance Imaging (MRI) and Confocol microscopic images using three-dimensional metamorphosis techniques. The technique is based on interpolation in the Fourier domains of two brain models and then transforming the interpolated results back. We introduce ICP algorithm to align brain models and use a proper filter to remove high frequencies while interpolating in low frequencies. The experiment result shows that alignment yields more satisfying result.