In this work, we propose a novel deep learning-based framework for the adaptive configuration of DMRS used for MIMO CCE. Specifically, a neural network architecture is proposed at the terminal side which based on statistical learning indicates to the network the preferred configuration of DMRS in time, frequency, and code domains for use in subsequent slots. The neural network in this work is trained offline through utilising the default DMRS configuration(s) from the network to assess the accuracy of the CCE process based on a practical minimum meansquare error (MMSE) channel estimation algorithm. The results presented are obtained through utilising synthetic 5G NR standards-compliant waveforms and channel models corresponding to a number of scenarios with different channel conditions and signal-to-noise ratios (SNRs). In order to prevent over-fitting, the synthetic data were divided into training and validation sets. The results in this paper highlight that the significant gains can be achieved in terms of reduced DMRS overhead improved CCE performance under same DMRS overhead, and in certain cases both, through applications of deep learning for adaptive DMRS configuration.