Detector-Stage Failures Account for Most of the Increase in the External Deficit of a Box-Prompted Glioma Segmentation Pipeline on BraTS-Africa: A Three-Arm Decomposition Across Two External Cohorts.
Authors
Affiliations (2)
Affiliations (2)
- Department of Computer Engineering, Ankara Yıldırım Beyazıt University, Ankara 06010, Türkiye.
- Department of Computer Engineering, Turkish Military Academy, National Defence University, Ankara 06420, Türkiye.
Abstract
<b>Background/Objectives:</b> When a box-prompted segmentation pipeline loses accuracy on new data, the detector proposing the box or a later stage may be responsible; this study measured which. <b>Methods:</b> A real-time detection transformer (RT-DETR) feeding the Segment Anything Model 2.1 was compared with a detector-free model (MedSAM3 with a low-rank adapter) on whole-tumor Dice in 158 patients with glioma: held-out Brain Tumor Segmentation (BraTS) 2020 cases (<i>n</i> = 24), the RHUH-GBM glioblastoma cohort (<i>n</i> = 39), and BraTS-Africa (<i>n</i> = 95, glioma presumed). Giving the segmenter the ground-truth box splits the Dice gap between the systems into a detector-stage term and a residual. <b>Results:</b> The detector-stage term was +0.006 Dice on held-out cases, +0.018 on RHUH-GBM, and +0.091 on BraTS-Africa. Its increase over the held-out value was +0.0854 [95% CI +0.0450, +0.1303] on BraTS-Africa (Holm-adjusted <i>p</i> < 0.001) and, after correction of a preprocessing fault, +0.0120 [-0.0062, +0.0295] on RHUH-GBM, which was not resolved. On BraTS-Africa, it came mainly from slices without a detector box. No residual change was resolved. <b>Conclusions:</b> On BraTS-Africa, the detector stage accounted for most of the increase in the pipeline's gap; on RHUH-GBM, no increase was resolved. The decomposition should be measured per cohort.