Adaptive Post-Processing Recovers Most of the Gap to nnU-Net v2 in Head and Neck GTV Segmentation: A Paired Three-Arm HECKTOR 2025 Benchmark
- 1Radiation Oncology Department, Hospital Universitario Puerta del Mar, Servicio Andaluz de Salud, Cadiz, Spain
Abstract
BackgroundAccurate delineation of the gross tumour volume (GTV) -- primary tumour (GTVp) and nodal disease (GTVn) -- on FDG-PET/CT is a critical step of head and neck radiotherapy planning. Comparisons between lightweight custom networks and the auto-configured nnU-Net v2 are usually reported as end-to-end pipelines, conflating the contribution of the network with that of the inference-time post-processing applied on top of it. We separated the two.
MethodsMiniUNet3D (custom 3D U-Net, 18.3 M parameters) and nnU-Net v2 (3d_fullres, 88.2 M parameters) were trained on the same 578 FDG-PET/CT cases (85/15 author-defined split of the HECKTOR 2025 Task 1 set, 8 centres) and evaluated on the same internal cohort. Three arms were compared pairwise: MiniUNet3D raw output at a fixed 0.5 threshold, MiniUNet3D with a locked adaptive post-processing pipeline, and nnU-Net v2. Comparisons used paired Wilcoxon tests with bootstrap confidence intervals, Bonferroni and Benjamini-Hochberg correction, and Cohens d; catastrophic failure (Dice <0.01) was compared with an exact McNemar test. Cases with an empty reference for a given target were excluded from that targets analysis (n = 98 GTVp, n = 93 GTVn).
ResultsWith post-processing matched off, nnU-Net v2 was superior: median GTVp Dice 0.799 versus 0.592 (mean difference -0.244, 95 % CI -0.300 to -0.191; d = -0.88) and GTVn 0.774 versus 0.598 (d = -0.82). Post-processing raised MiniUNet3D to 0.800 (GTVp) and 0.738 (GTVn), recovering 79 % of that difference. Post-processed, MiniUNet3D matched nnU-Net v2 on GTVp Dice (p = 0.113) but remained inferior on nodal disease after Bonferroni correction (Dice p = 0.041; surface Dice p = 0.049). Catastrophic GTVp failures were 25/98 raw, 8/98 post-processed and 1/98 for nnU-Net v2 (McNemar p = 0.016). Inference took 34 s versus 78 s per case on the same GPU.
ConclusionsPost-processing recovered most, but not all, of the difference between the two models, and it did not confer robustness: an eight-fold higher rate of empty contours on small primaries persisted, which is the more consequential difference for planning safety. Pipeline comparisons reported without a post-processing ablation risk attributing to a network what post-processing supplied.
HighlightsO_LIThree-arm paired benchmark on HECKTOR 2025 Task 1: a lightweight custom 3D U-Net (MiniUNet3D, 18.3 M parameters) with and without adaptive post-processing, against the auto-configured nnU-Net v2 (88 M parameters), on an identical training/validation split. C_LIO_LIWith post-processing matched off in both arms, nnU-Net v2 is decisively better: median GTVp DSC 0.799 vs 0.592 (paired mean -0.244; Cohens d = -0.88; p < 0.001 after Bonferroni). C_LIO_LIAdaptive post-processing recovers most of that gap: it adds +0.192 DSC to MiniUNet3D on GTVp, 79 % of the 0.244 model-level difference. Post-processing is thus slightly smaller than the model gap itself, but three to four times larger than what remains of it afterwards (0.051 - 0.062). C_LIO_LIThe gap is not fully closed. Even post-processed, nnU-Net v2 retains a small but Bonferroni-significant advantage on nodal disease (GTVn DSC p = 0.041; sDSC@2mm p = 0.049) and an eight-fold lower catastrophic failure rate (1/98 vs 8/98; McNemar exact p = 0.016). C_LIO_LIPractical implication: a team already committed to a lightweight custom network can recover most of what that choice costs at the inference stage, without retraining -- but post-processing does not buy robustness against catastrophic failure on small primaries. C_LI
Related
- Urinary stone segmentation: A computationally efficient yet consistently effective approach using image processing
- A Sentinel-2 benchmark dataset for deep-learning active-fire segmentation across 25 California wildfires, July 2020–August 2026
- AMF-U-Net: an adaptive multimodal fusion residual attention 3D U-Net for boundary-aware brain tumour segmentation
- Anatomically constrained hierarchical post-processing for pediatric brain tumor segmentation
- Calibration and processing sensitivity of time-embedded three-dimensional electrocardiographic descriptors: a two-dataset benchmark
BibTeX
@misc{silva2026adaptive,
title = {{Adaptive Post-Processing Recovers Most of the Gap to nnU-Net v2 in Head and Neck GTV Segmentation: A Paired Three-Arm HECKTOR 2025 Benchmark}},
author = {R. Oyarzun Silva and P. Hernandez Hernandez},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.28.26361649},
url = {https://www.medrxiv.org/content/10.64898/2026.08.28.26361649v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Adaptive Post-Processing Recovers Most of the Gap to nnU-Net v2 in Head and Neck GTV Segmentation: A Paired Three-Arm HECKTOR 2025 Benchmark
AU - R. Oyarzun Silva
AU - P. Hernandez Hernandez
PY - 2026
DA - 2026/08/28
PB - medRxiv
DO - 10.64898/2026.08.28.26361649
UR - https://www.medrxiv.org/content/10.64898/2026.08.28.26361649v1
AB - BackgroundAccurate delineation of the gross tumour volume (GTV) -- primary tumour (GTVp) and nodal disease (GTVn) -- on FDG-PET/CT is a critical step of head and neck radiotherapy planning. Comparisons between lightweight custom networks and the auto-configured nnU-Net v2 are usually reported as end-to-end pipelines, conflating the contribution of the network with that of the inference-time post-processing applied on top of it. We separated the two. MethodsMiniUNet3D (custom 3D U-Net, 18.3 M parameters) and nnU-Net v2 (3d_fullres, 88.2 M parameters) were trained on the same 578 FDG-PET/CT cases (85/15 author-defined split of the HECKTOR 2025 Task 1 set, 8 centres) and evaluated on the same internal cohort. Three arms were compared pairwise: MiniUNet3D raw output at a fixed 0.5 threshold, MiniUNet3D with a locked adaptive post-processing pipeline, and nnU-Net v2. Comparisons used paired Wilcoxon tests with bootstrap confidence intervals, Bonferroni and Benjamini-Hochberg correction, and Cohens d; catastrophic failure (Dice <0.01) was compared with an exact McNemar test. Cases with an empty reference for a given target were excluded from that targets analysis (n = 98 GTVp, n = 93 GTVn). ResultsWith post-processing matched off, nnU-Net v2 was superior: median GTVp Dice 0.799 versus 0.592 (mean difference -0.244, 95 % CI -0.300 to -0.191; d = -0.88) and GTVn 0.774 versus 0.598 (d = -0.82). Post-processing raised MiniUNet3D to 0.800 (GTVp) and 0.738 (GTVn), recovering 79 % of that difference. Post-processed, MiniUNet3D matched nnU-Net v2 on GTVp Dice (p = 0.113) but remained inferior on nodal disease after Bonferroni correction (Dice p = 0.041; surface Dice p = 0.049). Catastrophic GTVp failures were 25/98 raw, 8/98 post-processed and 1/98 for nnU-Net v2 (McNemar p = 0.016). Inference took 34 s versus 78 s per case on the same GPU. ConclusionsPost-processing recovered most, but not all, of the difference between the two models, and it did not confer robustness: an eight-fold higher rate of empty contours on small primaries persisted, which is the more consequential difference for planning safety. Pipeline comparisons reported without a post-processing ablation risk attributing to a network what post-processing supplied. HighlightsO_LIThree-arm paired benchmark on HECKTOR 2025 Task 1: a lightweight custom 3D U-Net (MiniUNet3D, 18.3 M parameters) with and without adaptive post-processing, against the auto-configured nnU-Net v2 (88 M parameters), on an identical training/validation split. C_LIO_LIWith post-processing matched off in both arms, nnU-Net v2 is decisively better: median GTVp DSC 0.799 vs 0.592 (paired mean -0.244; Cohens d = -0.88; p < 0.001 after Bonferroni). C_LIO_LIAdaptive post-processing recovers most of that gap: it adds +0.192 DSC to MiniUNet3D on GTVp, 79 % of the 0.244 model-level difference. Post-processing is thus slightly smaller than the model gap itself, but three to four times larger than what remains of it afterwards (0.051 - 0.062). C_LIO_LIThe gap is not fully closed. Even post-processed, nnU-Net v2 retains a small but Bonferroni-significant advantage on nodal disease (GTVn DSC p = 0.041; sDSC@2mm p = 0.049) and an eight-fold lower catastrophic failure rate (1/98 vs 8/98; McNemar exact p = 0.016). C_LIO_LIPractical implication: a team already committed to a lightweight custom network can recover most of what that choice costs at the inference stage, without retraining -- but post-processing does not buy robustness against catastrophic failure on small primaries. C_LI
ER -