The present invention relates to a method for predicting a fracture by evaluating the distribution of bone density and using the same, and more specifically, acquiring an image of a two-dimensional image through a dual energy X-ray absorption measurement method, and then evaluating and using the bone density distribution. And how to predict a fracture. A method for predicting a fracture using a DXA image according to the present invention comprises: a first step of obtaining a DXA image of a lumbar spine； A second step of reconstructing an internal bone density distribution map from the DXA image； A third step of extracting a distribution map of trabecular bone bone density from the reconstructed bone density distribution map； A fourth step of setting an ROI region by evaluating the direction of force applied to the inner bone from the extracted trabecular bone bone density distribution map； A fifth step of measuring an average value of a distribution map of trabecular bone density corresponding to the ROI region； And a sixth step of calculating a fracture risk by using the average value of the measured trabecular bone bone density. In the second step, the DXA image is preferably divided into 12,000 pixels in a matrix of 150*800 for one vertebra to reconstruct the bone density distribution map. In the third step, extracting the distribution of trabecular bone density is obtained by obtaining a standard deviation between neighboring cells and cells from the reconstructed bone density distribution, and then multiplying the standard deviation by a weight in a preset range. It is characterized by extracting the distribution of. In the fourth step, the ROI region is a region of 45° clockwise and counterclockwise with respect to a central vertical line in the distribution map of the trabecular bone bone density extracted in the third step. In the fifth step, the bone density weight is calculated by multiplying the trabecular bone bone density distribution value correspond
