Abstract

Pelvic lymph-node status dictates surgical and adjuvant therapy in gynecologic cancers. MRI-based assessment, however, is trapped by a fundamental dilemma: metastatic nodes frequently measure below the 10-mm threshold used to define suspicious enlargement, rendering size-based triage inherently unreliable. To overcome this, SLN-Net is proposed as a coarse-to-fine dual-sequence framework integrating contrast-enhanced T1-weighted (T1) and T2-weighted imaging (T2). The framework confines detection to anatomically constrained lymphatic-drainage territories, suppressing false positives from vessels and soft tissues, and subsequently applies a small-object-aware detector with a high-resolution P2 branch and bidirectional fusion to preserve the fragile signals of sub-centimeter nodes. On 418 patients with strict patient-level cross-validation, SLN-Net outperforms the best generic detector (YOLOv11) by 34.2% in mAP@0.5 (from 0.547 to 0.734) and 37.3% in precision (from 0.557 to 0.765), with a positive-node recall of 0.646. Ablation confirms these two design choices: region guidance alone raises mAP by 27.4% (from 0.547 to 0.697), and the small-object multiscale module further improves it to 0.734. On an independent external-center cohort (64 patients, 4,088 images) without fine-tuning, patient-level accuracy is 68.8%. With an average inference time of 20 seconds per patient (for a full MRI series of approximately 70 slices), SLN-Net shows potential as a prescreening tool for reducing missed small lymph nodes in clinical practice.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/CaPTion_025.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: Not Submitted

Link to Open Review

Open Review Page: https://openreview.net/profile?id=~Xin_Wang62

BibTex

@InProceedings{XuChe_SLNNet_MICCAISAT2026,
        author = { Xu, Chenchen AND Lu, Zhiyi AND Wang, Haotian AND Chen, Yabin AND Yao, Jun AND Huang, Minghao AND Qin, Fengying AND Liu, Guanyu AND Yu, Tianwei AND Wang, Yurui AND Deng, Renfu AND Yuan, Junhui AND Wen, Feng AND Wang, Huilan AND Wu, Jiesheng AND Wang, Guili AND Dong, Yue AND Wang, Xin},
        title = { { SLN-Net: A Dual-Sequence MRI Network for Detecting Small Lymph Nodes in Gynecologic Malignancies } },
        booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
        year = {2026},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 17278},
        month = {pending},
        page = {pending}
}


back to top