Measurement of magnetic signals generated from electric activity of myocardium provides useful information for the functional diagnosis of heart diseases. Key technical component of the magnetocardiography (MCG) technology is SQUID. To measure MCG signals with high signal-to-noise ratio, sensitive SQUID magnetic field sensors are needed. Present magnetic field sensors based on Nb SQUIDs have field sensitivity good enough to measure most of MCG signals. However, for accurate measurement of fine signal pattern or detection of local atrial fibrillation signals, we may need higher field sensitivity. In addition to field sensitivity, economic aspect of the SQUID system is also important. To simplify the SQUID readout electronics, the output voltage or flux-to-voltage transfer of SQUID should be large enough so that direct measurement of SQUID output can be done using room-temperature preamplifiers. Double relaxation oscillation SQUID (DROS), having about 10 times larger flux-to-voltage transfers than those of DC-SQUIDs, was shown to be a good choice to make the electronics compact. For effective cancellation of external noise inside a thin economic shielded room, first-order axial gradiometer with high balance, simple structure and long-baseline is needed. We developed a technology to make the axial gradiometer compact using direct bonding of superconductive wires between pickup coil and input coil. Conventional insert has mechanical support to hold the gradiometer array, and the dewar neck has equal diameter with the dewar bottom. Boiling of the liquid He can generate mechanical vibrations in the gradiometer array due to mechanical connection structure. Elimination of the mechanical support, and direct mounting of the gradiometer array into the dewar bottom can reduce the dewar neck diameter, resulting in the reduction of liquid He consumption.