.png){kind=icon}
.png){kind=icon}
Return
HIFU micro-damage detection method based on optical flow multi-parameter time-division analysis
.png){kind=icon}
PRE
AI
DOI：10.1016/j.bspc.2025.108412.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
High-intensity focused ultrasound (HIFU) technology has demonstrated significant clinical potential in the treatment of both benign and malignant tumors. However, existing ultrasound monitoring techniques suffer from low sensitivity and insufficient precision, leading to a reliance on subjective judgment during treatment. This study proposes a novel HIFU micro-damage detection method based on optical flow multi-parameter temporal analysis to address this issue. This method utilizes dynamic ultrasound monitoring data and develops a detection algorithm centered around an optical flow model aimed at accurately detecting small HIFU-induced damage in biological tissues. Specifically, the process involves first segmenting the preprocessed HIFU ultrasound image sequence into time periods and frame pairs. Next, a multi-resolution optical flow model is used to estimate the optical flow and extract motion information from the damaged target area. Multiple parameters, including the optical flow intensity, motion direction, and grayscale changes, are subsequently extracted through optical flow decomposition. Finally, a multi-parameter temporal analysis network is constructed to evaluate the damage and achieve precise identification of small damage regions. Verification through 56 sets of ex vivo pork experiments shows that this method achieves a good balance between accuracy (91.3%) and recall, with an F1 score of 81.4%. The method outperforms traditional approaches, such as grayscale detection, with an average of 24.1% improvement in precision and an AUC value of 0.843. In conclusion, this study significantly enhances the sensitivity and accuracy of HIFU micro-damage detection, providing new technological support for clinical treatments.
Keywords:
HIFU
optical flow
micro-damage detection
multi-parameter analysis
ultrasound monitoring
Journal
IF:
4.9
Papers:
9.8K
Citations:
2.4W