We jointly design the reconfigurable ring convolutional encoders and reconfigurable convolutional interleavers for the MPEG-4 medical image transmission system. Our objective is to maintain our desirable system performance due to the variation of channels with the lowest system complexity. In order to gain the performance in burst noise environment, we usually apply an interleaver by trading off with the system complexity. However, for a given block-fading signal interval, Ns, we observe that the system performance reaches a certain value regardless of the number of shift registers in the interleaver, N . By adjusting the number of states in the best ring convolutional encoder, St, we summarize the simulation results in terms of the peak signal to noise ratio (PSNR) and word error rate (WER) system performance with N ranging from 5 to 30. The overall average PSNRs are obtained between around 11 and 28 dB by assigning St to 4, 8, 16 and 32 states. Moreover, we are able to achieve the highest average PSNR improvement up to 3 dB.