Yoshiki Kawata PhD, PRESENTER: Nothing to Disclose

Abstract: HTML Purpose: To develop numerical criteria for classifying nodule density patterns that provides information with respect to nodule statuses. Methods and Materials: Pulmonary nodules on high-resolution CT images revealed several density distribution patterns depending the nodule status. From the visual assessment, density distribution patterns might be classified into some categories such as a nodule with ground-glass opacity alone, a nodule that has ground-glass opacity only at periphery, and solid nodule. In order to quantify CT density distribution inside each nodule, three-dimensional (3-D) image processing techniques were used in 158 patients for whom the final diagnosis was known (57 benign and 101 malignant cases) and whose initial nodule diameters were less than 20 mm. In the classification method, first, the nodule was volumetrically separated into three-layer parts so that each layer part had almost equality volume. Second, for each layer part, a CT density histogram was obtained and the three features, peak position, sign of skewness, and sign of kurtosis, were computed to characterize the histogram form. Finally, using combination the features, nodule types were classified into five classes, (A), (B), (C), (D), and (E). The (A) and (B) types respectively referred to the nodule with ground-glass opacity alone and one that has ground-glass opacity only at periphery. The other types, (C), (D), and (E) were considered as sub classes of the solid type. Results: The data set used was tracked during nine years from 1993 to 2002. The number of mortality cases was five. Among 158 cases, 35 involved (A) and (B) types. Among these types, malignancy was diagnosed in 28 (80%) as opposed to a 59% malignancy rate for other types, (C), (D), and (E). The malignancy rate for (B) type was 82%, and the rate for (A) type was 77%. Among 64 malignancies including all mortality cases with histological diagnosis such as stage and Noguchi type, all (A) and (B) types were diagnosed as stage I and included no mortality cases. The other types, (C), (D), and (E) less than 10 mm in diameter included no mortality cases. Conclusion: A classification method of pulmonary nodules was developed to quantify the internal density structure using high-resolution CT images. This method has highly potential usefulness to assess benignity and malignancy statuses.      

Kawata PhD, Y, A Classification Method of Pulmonary Nodules Using 3-D CT Images.  Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL. http://archive.rsna.org/2003/3108527.html