MPF-net: An effective framework for automated cobb angle estimation

Kailai Zhang, Nanfang Xu, Chenyi Guo, Ji Wu

Abstract: Highlights•We design a multi-task framework for joint learning of vertebra detection and landmark prediction, which can improve both performance.•We propose a novel proposal correlation module based on the relationship of vertebras, which can find the hidden vertebras effectively.•We design a feature fusion module to utilize the information of multi-view X-rays for better performance.•We achieve greater performance for Cobb angle measurement than previous works, so that our method can provide clinicians with automated and effective Cobb angle estimation.