CAJ | 학술논문

目的建立多层螺旋CT在Z轴空间分辨率无损条件下影像重组时最适重建间隔的数学模型与算法表达.方法根据高斯理论及信号抽样原理,推导出Z轴空间分辨率无损重建最适重建间隔的算法表达式,该式是有效层厚的函数.应用Siemens 64层螺旋CT与内耳扫描模式对多层螺旋CT层敏感度曲线(SSP)测试体模扫描,测试SSP和有效层厚,进行客观评价.分别以0.100、0.300、0.400、0.500 mm间隔对SSP进行傅立叶变换,获得Z轴调制传递函数(MTF).选择临床患者6例7耳(仅健侧颞骨),分别以0.100、0.300、0.400、0.500 mm间隔重建横断面图像,并依此进行冠状面MPR.由3名高年资影像诊断医师对9个高对比度微细解剖结构影像图进行软阅读盲法评分.以重建间隔为处理因素对临床评分数据行双因素方差分析,并对处理组数据行两两比较(Dunnettt检验).结果 (1)标称层厚0.600 mm的测试值(有效层厚)是0.665 mm,根据本研究中算法计算的最适重建间隔是0.296 mm(≈0.300 mm).(2)客观评价显示重建间隔0.100和0.300 mm的MTF曲线相重合且没有混叠,重建间隔0.400和0.500 mm的MTF曲线在中高频区域出现混叠.(3)不同重建间隔的颞骨冠状面MPR的临床评分差异有统计学意义(F=505.374,P＜0.01).以重建间隔0.100 mm为控制组进行处理组间的两两比较,与重建间隔0.300 mm相比差异无统计学意义(t=-0.222,P＞0.05),与重建间隔0.400和0.500 mm相比差异均有统计学意义(t值分别为-1.333、-15.889,P值分别＜0.05、＜0.01).结论客观评价和临床评价均与数学模型计算结果相符,数学模型和算法表达式可以作为影像重组时最适重建间隔的设置依据.
목적 건립다층라선CT재Z축공간분변솔무손조건하영상중조시최괄중건간격적수학모형여산법표체.방법 근거고사이론급신호추양원리,추도출Z축공간분변솔무손중건최괄중건간격적산법표체식,해식시유효층후적함수.응용Siemens 64층라선CT여내이소묘모식대다층라선CT층민감도곡선(SSP)측시체모소묘,측시SSP화유효층후,진행객관평개.분별이0.100、0.300、0.400、0.500 mm간격대SSP진행부립협변환,획득Z축조제전체함수(MTF).선택림상환자6례7이(부건측섭골),분별이0.100、0.300、0.400、0.500 mm간격중건횡단면도상,병의차진행관상면MPR.유3명고년자영상진단의사대9개고대비도미세해부결구영상도진행연열독맹법평분.이중건간격위처리인소대림상평분수거행쌍인소방차분석,병대처리조수거행량량비교(Dunnettt검험).결과 (1)표칭층후0.600 mm적측시치(유효층후)시0.665 mm,근거본연구중산법계산적최괄중건간격시0.296 mm(≈0.300 mm).(2)객관평개현시중건간격0.100화0.300 mm적MTF곡선상중합차몰유혼첩,중건간격0.400화0.500 mm적MTF곡선재중고빈구역출현혼첩.(3)불동중건간격적섭골관상면MPR적림상평분차이유통계학의의(F=505.374,P＜0.01).이중건간격0.100 mm위공제조진행처리조간적량량비교,여중건간격0.300 mm상비차이무통계학의의(t=-0.222,P＞0.05),여중건간격0.400화0.500 mm상비차이균유통계학의의(t치분별위-1.333、-15.889,P치분별＜0.05、＜0.01).결론 객관평개화림상평개균여수학모형계산결과상부,수학모형화산법표체식가이작위영상중조시최괄중건간격적설치의거.
Objective To study algorithm expression of the optimal reconstruction increment with which images could be reconstructed without loss of z-direction spatial resolution for multislice spiral CT.Methods Using Gauss function and signal sampling principle,an algorithm expression was deduced to calculate the optimal reconstruction increment with which images could be reconstructed without loss of zdirection spatial resolution.Spiral slice sensitivity profile (SSP) phantom was scanned using Somatom Sensation 64-slice spiral CT and temporal bone protocol as those used for clinic,axial images were reconstructed with slice thickness of 0.600 mm and increment of 0.100,0.300,0.400 and 0.500 mm respectively.Then SSPs and full width at half maximum (FWHM) were measured and modulation transfer functions were obtained by Fourier transfer from SSPs.Axial CT scan of 7 normal temporal bones in 6 patients were obtained by the same CT system and parameters as above.Coronal MPR images of temporal bone out of different reconstruction increment were obtained and the quality of reconstructed images were independently assessed by three senior radiologists using a four-point scale and blinded the information of reconstruction.Experimental data were processed and two-way ANOVA(in which Dunnett t test was selected for multiple comparisons) was performed with statistic software SPSS10.0.P ＜ 0.05 was considered as significant difference.Results (1) The measured FWHM of reconstruction slice thickness of 0.600 mm was 0.665 mm,so the optimal reconstruction increment calculated with the algorithm expression in this article was 0.296 mm(≈0.300 mm).Objective evaluation showed that there was obviously aliasing in high spatial frequency range of MTF curve when reconstruction increment was more than 0.300 mm,i.e.both 0.400 and 0.500 mm.Clinical scores of coronal MPR images of temporal bone reconstructed with different increment had significant difference(F =505.374,P ＜ 0.01).Treating reconstruction increment of 0.100 mm as a control group and comparing another groups against it,there was no significant difference for reconstruction increment of 0.300 mm (t =-0.222,P ＞ 0.05),while there were significant difference for 0.400 and 0.500 mm (t =-1.333、-15.889,P ＜ 0.05,P ＜ 0.01).Conclusion It was proved that the angorithm expression deduced in this article was proper because of the consistence between the calculated value and the results from objective and clinical evaluation.The optimal reconstruction increment without loss of z-direction spatial resolution is 0.445 times as large as FWHM for multislice CT.