CT Quantitative Diagnosis In Solitary Pulmonary Nodules

Objective To evaluate the value of the maximum likelihood method in differentiating the 4 kinds of SPN among peripheral lung carcinoma, benign tumors, tuberculomas, inflammatory nodules; to evaluate the value of multi-slice spiral CT perfusion imaging (CTPI) in differentiating SPN.Materials and Methods Sixty cases of SPN with pathologically or clinically proved were selected for studying objects, including 36 cases of peripheral carcinomas, 6 benign tumors, 8 tuberculomas, 10 inflammatory nodules. Analyzing 60 cases of SPN with the maximum likelihood method, the concrete steps of implement were included:⑴To avoid error caused by the small sample size, 455 patients with SPN(including 166 cases of peripheral lung cancer, 92 benign tumors, 98 tuberculomas, 59 inflammatory nodules) were collected from recent domestic three literatures, referring to similar practice in former document.⑵For a systematic statistic analysis ,10 features of SPN on CT were selected, in the application of the maximum likelihood method, transforming the occurrence probabilities for all the features of a nodule into scores, then building standards of discrimination.⑶Differentiating 60 SPN with standards already built. Two observers analyzed these SPNs with the traditional method collectively, without knowing the pathological or clinical diagnoses. Comparison was made on diagnostic results from the use of traditional method and the maximum likelihood method. Twenty-two cases of 60 SPNs were treated with multi-slice spiral CT perfusion imaging (CTPI) during the same period, including 11 peripheral carcinomas, 4 benign tumors, 4 tuberculomas, 3 inflammatory nodules. Diagnoses were made according the threshold (BF≥55ml/100ml/min, BV≥6ml/100ml and PS≥23ml/100ml/min) got from the study of perfusion imaging with 64-slice CT in solitary pulmonary spherical lesions, which were carried out during the same period.Results⑴The most probable CT signs for lung cancer were well-defined margin, lobulation, spiculation, and pleural indentation sign. The most probable CT signs for benign tumor were well-defined margin and calcification. The most probable CT signs for tuberculoma were well-defined margin, spiculation, pleural indentation sign, satellite, and calcification. The most probable CT signs for inflammatory nodules were spiculate protuberance, satellite.⑵The diagnostic accuracy with the maximum likelihood method of peripheral lung carcinoma, benign tumor, tuberculoma, inflammatory nodule were 80.5%(29/36), 83.5%(5/6), 75.0%(6/8), 80.0%(8/10). The mean accuracy was 80.0% (48/60). The diagnostic accuracy with the routine diagnostic method of peripheral lung carcinoma, benign tumor, tuberculoma, inflammatory nodule were 69.4%(25/36), 66.7%(4/6), 62.5%(5/8), 70.0%(7/10). The mean accuracy was 68.3% (41/60). Although the mean accuracy of former method was higher than the latter, there was no statistical difference(χ2=1.89, P=0.167>0.05).⑶The diagnostic accuracy was 86.4% of 22 SPNs treated with CTPI.Conclusions⑴The maximum likelihood method is a useful instrument for the statistical diagnosis of SPN, compared with the traditional method. Though without no statistically significance, it still can be applied to the daily routine study.⑵The CTPI study suggested there would be of significant value in differentiating malignant from benign SPN, if the perfusion parameter could be a diagnostic item of the maximum likelihood method.