High-field vertical standard (54 mm) bore superconducting magnets are widely used for NMR spectrometers. These magnets are very useful to MRI studies because the homogeneity and stability of the magnetic fields are excellent. However, their room temperature bore are usually made of Cu and eddy currents produced by the gradient field switching cause serious artifacts on the MR images. In particular, because echo-planar imaging requires intense readout gradients, solution of the eddy current problem is essential. We solved the problem using a proper size unshielded gradient coil set and reference scans, and succeeded in installing several EPI sequences for a 9.4 T superconducting magnet.We developed an MRI system using a 9.4 Tesla vertical standard bore superconducting magnet (JASTEC, Kobe, Japan), a home-built room temperature second-order shim coil set (50 mm o.d. and 40 mm i.d.), an unshielded gradient coil set (39 mm o.d. and 32 mm i.d.), and an eight-rung birdcage coil (19.5 mm o.d.) tuned to 400.4 MHz. The gradient coil elements were made from Cu rods as follows: coil patterns designed with the target field method were cut on the surface of the rods using a five-axis numerically controlled lathe; the trenches made by the lathe were filled with epoxy resin, and were finally bored to cylindrical shapes with 0.8 mm thickness. Three cylindrical gradient elements were assembled to a rigid gradient coil set using a vacuum impregnation technique. Figure 1 shows its picture, specifications, and eddy current properties measured in the magnet bore. A digital MRI transceiver (DTRX6, MRTechnology, Tsukuba, Japan) controlled by a PC controller was used for generation of pulse sequences [1]. One shot 2D EPI sequences with 1ms readout-gradient switching intervals, 64 ms signal readout time, and 80 ms spin-echo time were developed for a 15.36 mm square FOV (2 mm slice-thickness) with a 64 × 64 image matrix. Multishot 2D and 3D EPI sequences with 128 × 128, 256 × 256, and 256 × 256 × 16 image matrices were developed for a 15.36 mm square FOV (2 mm slice-thickness) and a 15.36 mm cube FOV. Before the EPI image acquisitions, the reference scans that phase encoding gradients were switched off were acquired to measure the peak positions and phases of the multiple gradient echo peaks. A water phantom consisting of 19 glass capillaries in a test tube (10 mm o.d and 9 mm i.d.) and a stem of celery were used for image evaluation.Figure 2 shows peak positions of the multiple gradient echoes in the switching intervals, phases of the gradient echo peaks, and differences of successive phases of the gradient echo peaks in the reference scans of the one-shot EPI sequences with Gx and Gy readout-gradients. These results were used to correct the positions along the readout gradients (resampling) and phases of the echoes. Figure 3 shows reconstructed EPI images with no correction, after resampling, and after both resampling and phase correction with Gx and Gy readout gradients. Figure 4 shows multishot EPI images with 128 × 128 and 256 × 256 pixels. The profiles of the enlarged images clearly demonstrated the spatial resolution of 120 and 60 mm. Figures 5(a) and (b) show 256 × 256 × 16 pixel 3D EPI images of the water phantom and a stem of celery. Slices near the central regions had no artifact but ghosting artifacts were observed for the slices near the edges.Figure 1 clearly shows that time constants of the eddy currents induced by the gradient switching are categorized into two groups; ~0.3 and ~2 ms. Because the readout gradient switching intervals were 1 ms, the eddy gradient fields with ~0.3 ms time constants affect the rise of the readout gradients but the effects stayed within the readout interval. In contrast, the eddy gradient fields with ~2 ms time constants affected later several echoes, which were observed as the decaying oscillation of the peak positions shown in Figs.2(a) and (d). The discontinuous change of the phases of the echo peaks are B0 shifts induced by the switching of the readout gradients. Although the EPI signals acquired in the magnet were affected by the eddy currents induced by the intense readout gradients, the effects were successfully corrected by the reference scan as shown in Figs.3-5. We therefore concluded that EPI was successfully installed in the standard and Cu bore vertical magnet using a proper size unshielded gradient coil set and the reference scans.