To explore the performance of improved watershed algorithm in processing magnetic resonance imaging (MRI) images and the effect of the processed images on the treatment of lumbar brucellar spondylitis (BS) with abscess by the posterior approach, the watershed algorithm was improved by adding constraints such as noise reduction and regional area attribute. 50 patients with abscessed lumbar disc herniation admitted to the hospital from January 2018 to January 2019 were selected, and all of them were examined by MRI. They were rolled into two groups in random. The treatment group (n = 25) accepted surgery with the aid of MRI images processed by the improved watershed algorithm, and the control group (Ctrl group) (n = 25) accepted surgery with the aid of unprocessed MRI images. The improved watershed algorithm can accurately segment the spine, and the segmentation results were relatively excellent. In contrast with the unprocessed MRI image, that processed by the improved watershed algorithm had a positive effect on the operation. In contrast with the Ctrl group, the visual analogue scale pain score (VAS), oxygen desaturation index (ODI), erythrocyte sedimentation rate (ESR), and high sensitivity C-reactive protein (CRP) were obviously lower ( p < 0.05 ). The improved watershed algorithm proposed performs better in MRI image processing and can effectively enhance the resolution of MRI images. At the same time, the posterior approach has a good effect in the treatment of lumbar BS with abscess and is worthy of clinical promotion.