Diagnostics of an ionospheric channel is based on an analysis of ionograms of vertical, slant, and return-slant sensing. This analysis calls for a solution of two general problems: 1) ionogram processing, that is, the correction and improvement of images and selection of points (moments of arrival) of a signal; 2) tracing tracks through the points for their subsequent referencing to specific propagation modes. To solve the first problem, statistical methods of image processing are used. To solve the second problem, the Houpfield method of artificial neuron nets (ANNs) is used. Because of the complex character of tracks against the intense background noise the modified rotor model has been used. Proper choice of the initial configuration has provided the fast convergence of the net. Ionograms recorded with chirp-zonde at the ISTP, Irkutsk, in 1987-1996 has been used to check the model. Analysis of the results has shown that this approach gives good results. It is promising for ionogram processing.