A method for automatically extracting preoperative image features of focused ultrasound ablation of uterine fibroids for re-pregnancy evaluation

Abstract

This invention relates to the field of biometric recognition software and image pattern analysis technology, and discloses an automatic extraction method for preoperative image features of uterine fibroids using focused ultrasound ablation for re-pregnancy assessment. The method includes: acquiring image voxel data of the target entity and calculating its geometric center; traversing candidate anchor points in a three-dimensional neighborhood space and identifying gradient maxima to determine the optimal sampling origin; emitting a spatial sampling vector from the optimal sampling origin to establish a radial sampling field; calculating the gradient and energy characteristics at interface attribute transitions; and fusing them after noise component stripping to generate a descriptor vector. This invention utilizes dynamic optimization of the sampling origin to achieve orthogonal alignment between the sampling path and the tissue interface, suppressing non-pathological diffusion of signal steps at the mechanistic level, and improving the accuracy and uniqueness of biometric descriptors in representing the physical integrity of tissues.

Images