CAJ | 학술논문

目的：探讨联合CT三维重建与 Fisher判别模式对术前肺结节良、恶性诊断的应用价值。方法回顾性分析40例孤立性肺结节(SPN)患者的CT资料,按病理及随访结果分为恶性肺结节25例,其中鳞癌4例、腺癌13例、肺泡癌4例、小细胞肺癌2例、大细胞癌1例、转移瘤1例；良性结节15例,其中结核球6例、错构瘤2例、非特异性炎性结节7例。并通过多平面重组(MPR)、曲面重组(CPR)、容积再现(VR)、最大密度投影(MIP)等三维重建技术对肺结节主要 CT 形态学特征进行评价,将三维数据分为良、恶性2组,以2组间每个征象有统计学意义的形态学征象为判别指标并行 Fisher判别,采用交叉核实法估计误判概率。结果三维重建所获得肺结节阳性征象明显比传统的二维图像多,其中三维重建所得 SPN CT征象的 Fisher判别公式为 Z=1.143X1+0.454X2+1.606X3-0.262X4+0.04X5+0.483X6+1.611X7-2.164。判别界值Zc为-0.516,Z>-0.516认为恶性结节可能性大,25例恶性结节中,有4例误判为良性；Z<-0.516认为良性结节可能性大,15例良性结节中,有2例误判为恶性。总误诊率为15%,准确率为85%。结论联合CT三维重建与 Fisher判别分析对肺结节良、恶性的判断,提高肺内结节的定性诊断,具有较高的临床价值。
목적：탐토연합CT삼유중건여 Fisher판별모식대술전폐결절량、악성진단적응용개치。방법회고성분석40례고립성폐결절(SPN)환자적CT자료,안병리급수방결과분위악성폐결절25례,기중린암4례、선암13례、폐포암4례、소세포폐암2례、대세포암1례、전이류1례；량성결절15례,기중결핵구6례、착구류2례、비특이성염성결절7례。병통과다평면중조(MPR)、곡면중조(CPR)、용적재현(VR)、최대밀도투영(MIP)등삼유중건기술대폐결절주요 CT 형태학특정진행평개,장삼유수거분위량、악성2조,이2조간매개정상유통계학의의적형태학정상위판별지표병행 Fisher판별,채용교차핵실법고계오판개솔。결과삼유중건소획득폐결절양성정상명현비전통적이유도상다,기중삼유중건소득 SPN CT정상적 Fisher판별공식위 Z=1.143X1+0.454X2+1.606X3-0.262X4+0.04X5+0.483X6+1.611X7-2.164。판별계치Zc위-0.516,Z>-0.516인위악성결절가능성대,25례악성결절중,유4례오판위량성；Z<-0.516인위량성결절가능성대,15례량성결절중,유2례오판위악성。총오진솔위15%,준학솔위85%。결론연합CT삼유중건여 Fisher판별분석대폐결절량、악성적판단,제고폐내결절적정성진단,구유교고적림상개치。
Objective To assess the dignosis value of CT three-dimensional reconstruction with Fisher discriminant model in small solitary pulmonary nodules before operation.Methods CT data of 40 cases with SPN were retrospectively analyzed and divided into into malignant pulmonary nodules (25 cases),squamous cell carcinoma (4 cases),adenocarcinoma (13 cases),lung cancer (4 ca-ses),small cell lung cancer (2 cases),large cell carcinoma (1 case),metastases tumor (1 case),benign nodules (1 5 cases,6 cases of tuberculosis,2 cases of hamartoma,and 7 cases of non-specific inflammatory nodules)by pathology and follow-up results.The CT features of pulmonary nodules were evaluated through multi-planar reformation (MPR),curved-planar reformation (CPR),volume rendering (VR),maximum intensity proj ection (MIP)and other three-dimensional reconstruction.The three-dimensional data were divided into benign and malignant groups.In each of the two groups,the significant signs of morphological signs of discrimination indicators were adminstrated Fisher discriminant,and the probalitiy of false positives were estimated using cross-validation method. Results The positive features of pulmonary nodules in there-dismensional images were much more than in two-dimensional images. Fisher discriminant formula of solitary pulmonary nodules in three-dimensional images was Z=1.143X1 + 0.454X2+1.606X3-0.262X4+0.04X5+0.483X6+1.611X7-2.164.Discriminant boundary value Zc was-0.516.When Zcgreater than -0.516,nodules were proneed to considere as malignant nodules.In 25 cases of malignant nodules,4 cases mistook for benign.When Zc less than -0.516,nodules were proneed to considere as benign nod-ules.In 1 5 benign nodules,2 cases mistook for malignant.The total misdiagnosis and accuracy rate were 15 % and 85% respec-tively.Conclusion CT three-dimensional reconstruction combined with Fisher discriminant model have a high clinical value in dif-fereiating diagonsis of pulmonary nodules were proneed to considere as malignant nodules.In 25 cases of malignant nodules,4 cases mistook for benign.When Zc less than -0.516,nodules were proneed to considere as benign nodules. In 15 benign nodules,2 cases mistook for malignant.The total misdiagnosis and accuracy rate were 15 % and 85% respec-tively.Conclusion CT three-dimensional reconstruction combined with Fisher discriminant model have a high clinical value in differeiating diagonsis of pulmonary nodules.