A method of matching nodes between MRI and pathology with MRI-based 3D node map in rectal cancer.

Abstract

Purpose: To propose a node-by-node matching method between MRI and pathology with 3D node maps based on preoperative MRI for rectal cancer patients to improve the yet unsatisfactory diagnostic performance of nodal status in rectal cancer.

Methods: This methodological study prospectively enrolled consecutive participants with rectal cancer who underwent preoperative MRI and radical surgery from December 2021 to August 2023. All nodes with short-axis diameters of ≥ 3 mm within the mesorectum were regarded as target nodes and were localized in three directions based on the positional relationship on MRI and drawn on a node map with the primary tumor as the main reference, which was used as a template for node-by-node matching with pathological evaluation. Patient and nodal-level analyses were performed to investigate factors affecting the matching accuracy.

Results: 545 participants were included, of whom 253 received direct surgery and 292 received surgery after neoadjuvant therapy (NAT). In participants who underwent direct surgery, 1782 target nodes were identified on MRI, of which 1302 nodes (73%) achieved matching with pathology, with 1018 benign and 284 metastatic. In participants who underwent surgery after NAT, 1277 target nodes were identified and 918 nodes (72%) achieved matching, of which 689 were benign and 229 were metastatic. Advanced disease and proximity to primary tumor resulted in matching difficulties.

Conclusion: An easy-to-use and reliable method of node-by-node matching between MRI and pathology with 3D node map based on preoperative MRI was constructed for rectal cancer, which provided reliable node-based ground-truth labels for further radiological studies.