Atlas-Guided Attribution Ranking Reveals Localization Gaps in C2 Fracture Classifiers: 16 Subregions and 101 Confirmed Cases With a Macro Dice of 0.88 for Five Groups in Scientific Reports, Published November 13
Scientific Reports published a study of localization in second cervical vertebra fracture classifiers on November 13, 2026. The paper, by Hu, Lin, Rohringer and co-authors, carries digital object identifier 10.1038/s41598-026-80027-w, was received on August 5, 2026 and accepted on November 4, 2026. The study reports an evaluation of an imaging analysis framework only and makes no individual treatment recommendation. The paper's starting point is that binary C2 fracture classifiers can identify whether a fracture is present but do not show which part of C2 is fractured. The team combined detailed C2-specific multi-atlas anatomy with classifier attribution maps to produce regional rankings, without requiring additional model training. The atlases came from three CT examinations from the RSNA 2022 Cervical Spine Challenge, manually annotated into 16 C2 subregions. The primary localization analysis included 97 examinations classified as positive by at least one classifier, and the secondary detection-to-localization analysis included all 101 confirmed C2 fracture cases. The dataset is publicly available and fully deidentified. Three-atlas leave-one-out evaluation of label propagation yielded macro Dice scores of 0.88 for five anatomical groups and 0.77 for 16 subregions, with weaker agreement for small articular facets. The authors declare that no funds, grants or other support were received for this study.
(Source: Scientific)
Source: Scientific
Published by GSM · USA (original Chinese edition)
Photo: 图：颈椎的X光正位片。Staff at the Department of Radiology，公共领域，经 Wikimedia Commons。 / Photo: Staff at the Department of Radiology, public domain, via Wikimedia Commons.
