Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1721
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-17T18:13:14.054784Z
status: published
lens: g_space
sector: health
headline: Peritumoral ultrasound-driven multimodal AI model for predicting lymphovascular space invasion and prognosis in cervical cancer
dek: Purpose Accurate preoperative assessment of lymphovascular space invasion (LVSI) in cervical cancer (CC) remains challenging despite its importance for treatment planning and risk stratification. This study aimed to develop and evaluate a hybrid model integrating peritumoral ultrasound radiomics features, deep learning features, and clinical data for preoperative prediction of LVSI in CC. Additionally, we assessed the model's prognostic value for disease-free survival (DFS) and explored LVSI-associated molecular…
gain_title: A hybrid model combining peritumoral ultrasound radiomics, deep learning scores and clinical data predicted lymphovascular space invasion in a prospective cohort better than single-modality models.
problem_title: (none)
trace_subject: (none)
gain_reading: A hybrid model combining peritumoral ultrasound radiomics, deep learning scores and clinical data predicted lymphovascular space invasion in a prospective cohort better than single-modality models.
gain_evidence: The hybrid model achieved the highest performance (AUC: 0.79, 95%CI: 0.66-0.92; sensitivity: 0.85, 95%CI: 0.71-0.98) in the prospective testing set, outperforming the single-modality models.
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built a model that analyzes ultrasound images of cervical tumors and the tissue around them, merging handcrafted radiomics features, transfer-learning deep features and clinical factors to estimate whether lymphovascular space invasion is present before surgery. The work used 292 women split across training, internal validation and prospective testing sets.

Preoperative assessment of this invasion is hard yet matters for treatment planning and risk stratification, so a noninvasive imaging tool could eventually support those decisions. The prospective cohort was small at 51 patients, the survival analysis was exploratory, and the authors say larger multicenter validation is needed before any clinical use.
limitation: The authors state that further validation in larger independent multicenter cohorts is needed before clinical application, and the prospective test set was small with a wide confidence interval.
tag: Evidence-backed gain
key_points: The study included a retrospective cohort of 241 women and a prospective cohort of 51 women, for 292 women in total. | The hybrid model reached an AUC of 0.79 and sensitivity of 0.85 in the prospective testing set. | Exploratory survival analysis showed shorter disease-free survival in the high-risk group. | TCGA data showed Wnt/β-catenin enrichment in LVSI-positive tumors, giving biological context.
rundown: Radiomics features were taken from tumor and peritumoral regions of interest on ultrasound images. LASSO regression and machine learning classifiers produced a Rad-score, while pre-trained convolutional neural networks produced a separate DL-score.

The hybrid model integrated clinical factors with both scores. The authors evaluated it by AUC and examined prognostic value with Kaplan-Meier analysis and interpretability with SHAP.

The patients were divided into 188 in the training set, 53 in internal validation and 51 in prospective testing. The authors present the biological pathway findings as complementary context rather than proof of mechanism.
sources:
- peer_reviewed | Abdominal Radiology | https://doi.org/10.1007/s00261-026-05804-7 | 2026-10-08
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

b367ca28aa8a417974a2cc264e95ca47d99ad2c595293576a04f867bc9a05d2b

previous

0000000000000000000000000000000000000000000000000000000000000000