Computer diagnosis of the roentgenogram is regarded as a typical two-dimensional pattern recognition problem. As the first step to approach the problem, we investigated the characteristics of the density distribution of the chest roentgenograms. Then we studied the method of automatic recognition of rib boundaries.
As the next step, we studied in this paper the method of automatic recognition of abnormal shadows in chest roentgenogram.
First, simple models of abnormal shadows in chest roentgenogram consisting of (1) some irregular figures whose uniform densities are known and whose shapes, sizes and positions are unknown, and (2) additive Gaussian noise with 0 mean and known variance, are assumed and recognition method is derived. Only outline of the method is stated here.
Next, this method is applied to the recognition of abnormal shadows in the real chest roentgenogram. In this experiment, input patterns (roentgenograms) are processed according to the following procedure : (i) measurement and sampling of film density, (ii) recognition of boundaries of dorsal portion of ribs, (iii) subtraction of density corresponding to dorsal portion of ribs, (iv) equalization of average density level in the horizontal direction, (v) recognition of boundaries of ventral portion of ribs, (vi) subtraction of density corresponding to ventral portion of ribs, (vii) recognition of abnormal shadows (tuberculous lesions).
Results of the experiment are stated here in detail. They show that the method used in this paper was proved useful for detection of some kinds of abnormal shadows in chest roentgenograms.
View full abstract