前列腺向膀胱內膨出之超音波影像分析

藉由電腦視覺以及醫療資訊兩個領域的結合，傳統超音波影像在醫療領域中很多都藉由人為判讀，但本次研究希望藉由電腦視覺的影像分析技術，提出一套可公式化的方法來處理超音波影像並提出判讀超音波影像的方法，研究主題主要針對膀胱及攝護腺的凹陷指數來判讀前列腺向膀胱內膨出的狀態，我們主要使用的識別方法包含:雜訊濾除、邊緣偵測、動態閥值、形態學、最短距離評估來做處理。並使用PCA距離測量及中心點垂直距離等方法來做量測標準。我們藉由濾除雜訊，找尋相對位子，對膀胱輪廓的規則整理，閥值比較以及分割合併還有橢圓近似等方法，能夠推斷出膀胱超音波影像的凹陷病變指數(IPP)並藉由橢圓近似還原出膀胱原本該有的樣子。

關鍵字

膀胱凹陷；攝護腺病變特徵；前列腺向膀胱內膨出；超音波影像；醫療影像處理

並列摘要

By combining computer vision and medical information, traditional ultrasound images in the medical field are often interpreted manually. However, in this study, we aim to propose a formulaic approach for processing ultrasound images and present a method for interpreting ultrasound images through computer vision techniques. The research focuses on the depression index of the bladder and prostate to assess the condition of the prostate protruding into the bladder. The identification methods used primarily include noise elimination, edge detection, dynamic threshold, morphology, and shortest distance evaluation. Additionally, we employ PCA distance measurement and vertical distance from the centroid as measurement standards. Through the elimination of noise, determination of relative positions, regularization of bladder contours, threshold comparison, segmentation and merging, and elliptical approximation, we can infer the intravesical prostatic protrusion (IPP) of bladder ultrasound images. By utilizing elliptical approximation, we aim to reconstruct the original appearance of the bladder from the ultrasound images.

並列關鍵字

Bladder indentation ； features of prostatic lesions ； Intravesical Prostatic Protrusion (IPP) ； ultrasound image analysis ； medical image processing

參考文獻

[1] Rafael C. Gonzalez • Richard E. Woods "Digital Image Processing" 4E, pp. 328-332

Google Scholar

[2] Peixuan Zhang & Fang Li(2014)" A New Adaptive Weighted Mean Filter for Removing Salt-and-Pepper Noise" IEEE Signal Processing Letters. pp.1280 - 1283