Hierarchical state space partitioning with a network self-organising map for the recognition of ST-T segment changes

The problem of maximising the performance of ST-T segment automatic recogni tion for ischaemia detection is a difficult pattern classification problem. The paper proposes the network self-organising map (NetSOM) model as an en hancement to the Kohonen self-organised map (SOM) model. This model is capa ble of effectively decomposing complex large-scale pattern classification p roblems into a number of partitions, each of which is more manageable with a local classification device. The NetSOM attempts to generalise the regula risation and ordering potential of the basic SOM from the space of vectors to the space of approximating functions. It becomes a device for the orderi ng of local experts (i.e, independent neural networks) over its lattice of neurons and for their selection and co-ordination, Each local expert is an independent neural network that is trained and activated under the control of the NetSOM. This method is evaluated with examples from the European ST- T database. The first results obtained after the application of NetSOM to S T-T segment change recognition show a significant improvement in the perfor mance compared with that obtained with monolithic approaches, i.e. with sin gle network types. The basic SOM model has attained an average ischaemic be at sensitivity of 73.6% and an average ischaemic bear predictivity of 68.3% . The work reports and discusses the improvements that have been obtained f rom the implementation of a NetSOM classification system with both multilay er perceptrons and radial basis function (RBF) networks as local experts fo r the ST-T segment change problem. Specifically, the NetSOM with multilayer perceptrons (radial basis functions) as local experts has improved the res ults over the basic SOM to an average ischaemic beat sensitivity of 75.9% ( 77.7%) and an average ischaemic beat predictivity of 72.5% (74.1%).