The radial overlap correction <math><msub><mi>δ</mi><mi>RO</mi></msub></math> is reexamined for 30 superallowed <math><mrow><msup><mn>0</mn><mo>+</mo></msup><mo>→</mo><msup><mn>0</mn><mo>+</mo></msup></mrow></math> nuclear <math><mi>β</mi></math> decays using the shell model with Hartree-Fock (HF) radial wave functions. Our mean-field calculation is based on the effective Skyrme interaction including Coulomb, charge-symmetry-breaking (CSB), and charge-independence-breaking (CIB) terms. In addition, the electromagnetic corrections such as those due to the gradient of charge density, vacuum polarization, Coulomb spin orbit, and finite size of nucleons are also considered. In order to avoid the spurious isospin mixing, the local equivalent potential is constructed from the solution of a HF calculation for the <math><mrow><mi>Z</mi><mo>=</mo><mi>N</mi></mrow></math> nucleus with charge-dependent forces neglected. Then the obtained mean field is solved noniteratively for the parent and daughter nuclei. It turns out that the CIB term has no significant impact on <math><msub><mi>δ</mi><mi>RO</mi></msub></math> throughout the mass range between 10 and 74. On the other hand, the CSB term makes <math><msub><mi>δ</mi><mi>RO</mi></msub></math> increase systematically between 10% and 30%. Similarly, the gradient density leads to a further increase of <math><msub><mi>δ</mi><mi>RO</mi></msub></math> between 2% and 14%, while other estimated electromagnetic corrections are negligible. The effect of the suppression of isospin spuriosity is somewhat complicated. In general, it produces a significantly larger value of <math><msub><mi>δ</mi><mi>RO</mi></msub></math>; however, there are a few cases for which <math><msub><mi>δ</mi><mi>RO</mi></msub></math> is mostly unaffected or even reduced, especially the even-even emitters in the light mass region. All these improvements partly explain the long-standing discrepancy between the correction values obtained with Woods-Saxon (WS) and HF radial wave functions. Nevertheless, the remaining discrepancy is still significant, except for the emitters with <math><mrow><mi>A</mi><mo>≤</mo><mn>26</mn></mrow></math>. In addition, the local odd-even staggering present in the results obtained with the phenomenological potential [J. C. Hardy and I. S. Towner, Phys. Rev. C 102, 045501 (2020); L. Xayavong and N. A. Smirnova, Phys. Rev. C 97, 024324 (2018)] is not well reproduced by our calculations. This subsequently leads to a large difference in the predicted mirror <math><mrow><mi>f</mi><mi>t</mi></mrow></math> ratios. These problems might relate to the errors of the calculated charge radii as well as the deformation and other correlation effects on the data of separation energies used for constraining the asymptotic radial wave functions.< Réduire