During prostate needle insertion, the gland and therefore targets move resulting in targeting inaccuracy. To compensate for this error, in a previous study we proposed robotic bevel-tip needle steering under live MRI. One important drawback is that tracking a flexible (20G) needle under real-time MRI is inaccurate due to the needle artifact and the image update frequency is relatively slow (>200 ms). As an alternative solution, we embedded some FBG strain sensors along the needle shaft at certain locations and reconstruct the needle shape from this information. Compared to the previous study on this topic by others, the following development have been done: 1) this study is extended to needle of higher gauge (thinner), 2) we guarantee below 0.5 mm accuracy at the needle tip for all insertion depth, 3) we extended it to bevel-tip needles (through modelling and mathematical formulation), 4) we visualized the 3D shape of the needle in 3D Slicer which is a commonly used planning and navigation software for prostate interventions, 5) our needle tracking is real-time.