Abstract: Accurate segmentation of liver CT images can provide an objective basis for liver disease research and early diagnosis, and it is also an important traceable project in medical metrology. This paper introduces an improved automatic segmentation method for liver CT images using a region growing approach, detailing the processing steps and implementation scheme. In the improved segmentation method, noise is first removed and image information is enhanced through threshold segmentation and nonlinear mapping. This method was used to segment nine randomly selected abdominal CT image sequences and compare the results with the threshold method. The Quasi-Monte Carlo (QMC) method is employed for seed point selection and to enhance the region growing conditions. Finally, morphological methods combined with the Canny operator and Flood fill algorithm are used for post-processing to smooth the liver contour and segmentation results. The results show that compared to the threshold method, the region growing method achieves a segmentation accuracy of up to 91.89%, allowing for more precise liver segmentation. Furthermore, the volume overlap error (VOE) and relative volume difference (RVD) statistics of the segmentation results also verify the effectiveness and stability of this method, indicating that it can be effectively applied in liver CT image segmentation and provide reliable support for the diagnosis of liver diseases in clinical medicine.