Tumor Classification via Subregion Combinations of Cortical L5PC6 Pyramidal Neurons

In this study, it was aimed to classify benign and malignant tumor samples by utilizing different subregions of a cortical L5PC6 pyramidal neuron. A spike-based learning method was applied to a real dataset, and a decision mechanism based on action potential generation was established by training a separate neuron for each class. During the training process, various types of synaptic connections and spike encoding schemes were evaluated and intensity coding yielded the highest performance for the pyramidal neuron. Regional analyses revealed that the basilar region particularly stood out in terms of classification success. The results demonstrate that the morphological and biophysical diversity of pyramidal neurons provides a significant advantage in classification tasks.