X’tal cube is the cubic depth of interaction (DOI) PET detector which our research group developed. In this work, aiming to get higher sensitivity, we developed the long rectangular shape X’tal cube (long-XC) by extending the cubic X’tal cube structure in one direction. We verified performance of this long-XC and also studied detector parameters for optimization. The same as the X’tal cube, the crystal block of the long-XC is composed of a 3D array of cubic scintillation crystal elements. Reflectors are not inserted between these crystal elements. The scintillation light then spreads without being obstructed by reflectors and is detected by multiple numbers of the multi-pixel photon counters (MPPCs) coupled on all six sides of the crystal block. For crystal element identification, a simple Anger-type calculation is used. In this study, we arranged 3.0 mm $\times 3.0$ mm $\times 3.0$ mm LGSO crystal elements in a $6\times 6\times 14$ array for the long-XC. In a previous study, we had already confirmed that for the X’tal cube consisting of a $6\times 6\times 6$ array of the same crystal elements and 54 MPPCs, identification of all 216 crystal elements was possible and the average energy resolution for all the elements was about 11 %. The long-XC contains more than twice the number of the crystal elements but less than twice the number of the MPPCs compared to the previous X’tal cube. The detector parameters investigated with the long-XC were: the number of MPPCs on both sides in the extended direction (edge MPPCs); the MPPC type, the MPPCs of $25\,\,\mu \text {m}\,\,\times 25\,\,\mu \text {m}$ or $50\,\,\mu \text {m}\,\,\times 50\,\,\mu \text {m}$ pixel sizes; the material between the crystal elements, an air gap or an optical glue having a closer refractive index to that of LGSO than air has; and the MPPC signals used in the Anger-type calculation. Results of the crystal element identification performance showed that reducing the number of the edge MPPCs caused performance degradation only at the part near the edge. For the MPPC type, the $50~\mu \text {m}$ type was better than the $25~\mu \text {m}$ type, and for the material, air was much better than the optical glue. We found that the choice of MPPC signals for the Anger-type calculation was effective in the optical glue condition. For the long-XC in the air gap condition and using the $50~\mu \text {m}$ type MPPCs, we observed it had good performance and there was no significant degradation at the central part which is far from the edge MPPCs. For irradiation of 662 keV gamma-rays, we measured approximately 11 - 13 % energy resolution for each crystal element and there was only a small difference in light outputs between crystal elements at the central part and at the edges. These results at the central part suggested the possibility of further extension of the long-XC for higher sensitivity.