Automatic LV Localization and Short-Axis Plane Estimation from Arbitrary CMR Slice

2026-07-31

AI Summary

This work addresses the challenge of accurately and continuously modeling the three-dimensional orientation of the left ventricle (LV) from a single cardiac magnetic resonance (CMR) slice, a task where existing methods suffer from limited generalizability and inadequate representation. The authors propose a novel paradigm that jointly performs LV localization and 3D short-axis plane orientation estimation by introducing a Polar-Coupled Circular (PCC) embedding, which enables continuous and unambiguous orientation modeling—overcoming limitations of conventional discrete classification or geometric intersection approaches. Additionally, they establish a scalable evaluation benchmark based on automated slice sampling from volumetric data. Evaluated across four datasets, the method achieves an average mIoU of 86.18% and a mean angular deviation of only 3.39°, significantly advancing the accuracy of single-slice LV orientation estimation.

Abstract

Accurate estimation of left ventricular (LV) orientation is essential for cardiac magnetic resonance (CMR) imaging and downstream analysis. Existing methods typically formulate orientation recognition as discrete view classification or rely on multi-slice geometric intersection, limiting their ability to model continuous 3D orientation and generalize across arbitrary slices. This work introduces a novel paradigm: Joint LV localization and 3D orientation estimation from a single CMR slice. To investigate this setting, representative orientation-aware detection frameworks are adapted to the CMR domain, and their limitations are analyzed. Upon that, we propose the Polar-Coupled Circular (PCC) embedding that provides a continuous and unambiguous orientation representation to address the limitations. Meanwhile, a scalable benchmark is constructed through automatic slice sampling from volumetric CMR segmentation datasets. Extensive experiments on four datasets demonstrate strong performance, achieving an average mIoU of 86.18% and an average angle deviation of 3.39°. This study establishes a new task setting for single-slice LV orientation modeling and provides a geometry-consistent framework for spatially informed CMR analysis. Code is available at https://github.com/yuyi1005/cmr-3d-ood.

Problem

• left ventricular orientation
• cardiac MRI
• single-slice analysis
• 3D orientation estimation
• arbitrary slice

Innovation

• Polar-Coupled Circular embedding
• single-slice CMR
• 3D orientation estimation
• left ventricular localization
• geometry-consistent representation