This paper presents Bit-Interleaved Coded Modulation metrics for joint estimation detection using training or reference signal transmission strategies for short to long block length channels. We show that it is possible to enhance the performance and sensitivity through joint detection-estimation compared to standard receivers, especially when the channel state information is unknown and the density of the training dimensions is low. The performance analysis makes use of a full 5G transmitter and receiver chains for both Polar and LDPC coded transmissions paired with BPSK/QPSK modulation schemes. We consider transmissions where reference signals are interleaved with data and both are transmitted over a small number of OFDM symbols so that near-perfect channel estimation cannot be achieved. This is particularly adapted to mini-slot transmissions for ultra-reliable, low-latency communications (URLLC) or for short packet random access use cases. We characterize the performance for up to eight receiving antennas in order to determine the performance gain offered by the proposed BICM detection in realistic base station receiver scenarios. Our findings demonstrate that when the detection windows used in the metric units is on the order of four modulated symbols the proposed BICM metrics can be used to achieve detection performance that is close to that of a coherent receiver with perfect channel state information for both polar and LDPC coded configurations. Furthermore, we show that for transmissions with low DMRS density, a good trade-off can be achieved in terms of additional coding gain and improved channel estimation quality by adaptive DMRS power adjustment.