Sara

we had the same issue. The implementation of nd-bhatta method is faulty. we reverted to 'bhatta' and that is fine, i.e. reasonable number of clusters.

yes, ours is MEA with 200 per electrode distancing in the probe design and I chose 100 for the radius.

True, but the hyper parameters are a problem if the number of spikes is really low (~ 100) with anything on the scale of thousands, the performance is robust to...