Automatic Bone Division and Lesion Detection by Using SPECT Whole Body Bone Scan ImagesAutomatic Bone Division and Lesion Detection by Using SPECT Whole Body Bone Scan Images

碩士 === 慈濟大學 === 醫學資訊學系碩士班 === 101 === Whole body bone scan is a common clinical routine investigation in nuclear medicine. The radioactivity substance, Tc-99m, marks on MDP and is intravenously injected into the bloodstream of a patient to help monitor this reaction. In bone lesions, more radioactiv...

Full description

Bibliographic Details
Main Authors: Pei-Ching Jang, 張沛晴
Other Authors: Sheng-Fang Huang
Format: Others
Language:zh-TW
Published: 2013
Online Access:http://ndltd.ncl.edu.tw/handle/18618674407167140325
Description
Summary:碩士 === 慈濟大學 === 醫學資訊學系碩士班 === 101 === Whole body bone scan is a common clinical routine investigation in nuclear medicine. The radioactivity substance, Tc-99m, marks on MDP and is intravenously injected into the bloodstream of a patient to help monitor this reaction. In bone lesions, more radioactive material will be seen, where metastases appear as areas of increased uptake. At present, bone scan imaging includes two modalities. One is called planar bone scan; the other is called SPECT (Single Photon Emission Computes Tomography). Single photon emission computed tomography (SPECT) with Tc-99m MDP supports three-dimensional (3D) data acquisition without superposed images of body structures. This study presents a bone division method and a quantitative scheme to detect the findings of possible abnormalities. This study consists of the following three phases. First, we use 3D region growing algorithm to remove the bladder region for avoiding over-enhanced effect. In addition, a modified histogram by adopting global contrast enhancement method is used to separate bone from background and soft tissues. Next, a bone division procedure is followed to separate skeletal structures into five sections and to remove knees and elbows. Finally, lesions in each bone section are detected according to an adaptive threshold value and symmetricity features. In this study, the datasets include 73 patients who received Tc-99m MDP whole body bone SPECT scan. In total 365 bone regions, there are 28 bone sections of bone lesions. The method shows 92.85% sensitivity, and the average of 1.25 false negative of each patient.