Course Content
Validating an isogenic tumor model for PSMA heterogeneity using [68Ga]-Ga-PSMA-11 microPET/CT
0/2
Novel non-steroidal fluorine-18 labeled PET tracer for imaging androgen receptors in a preclinical prostate cancer model
0/2
PET Imaging, Biodistribution and In Vitro Stability Assessment of [134Ce]Ce-RPS-088: A Versatile Surrogate for Non-Invasive Dose Calculation of [225Ac]Ac-RPS-088 in Prostate Cancer?
0/2
Dual-modality nuclear imaging: Development and in vivo evaluation of a novel Sc-44/ Lu chelation platform
0/2
Evaluating Theranostic Targets for Endometrial Carcinoma
0/2
The Deep Learning Radiomics Nomogram Helps to Evaluate the Lymph Node Status in Cervical Adenocarcinoma/Adenosquamous Carcinoma
0/2
Targeted Alpha and Beta Therapy with Sacituzumab-Govitecan: A TROP-2 targeted Theranostic Approach in a High-Grade Serous Ovarian Adenocarcinoma Mouse Model
0/2
Advancements in Prostate and Reproductive Health
About Lesson
Abstract Body:

Background Cervical cancer is the fourth most common cancer in women worldwide and the leading cause of cancer death in developing countries, accounting for about 85% of cases [1]. Adenocarcinoma (AC) and adenosquamous carcinoma (ASC) account for approximately 20-25% of cervical cancers and have exhibited a rise in morbidity and mortality over the past decades, particularly in young females [2]. The accurate assessment of lymph node metastasis (LNM) can facilitate clinical decision-making on radiotherapy or radical hysterectomy (RH), and prognosis evaluation in cervical AC/ASC [3-4].

Objectives Given the distinct biological properties, rising morbidity and mortality of AC and ASC, the objective of this study was to construct a DL radiomics nomogram (DLRN) for the accurate evaluation of LNM in cervical AC and ASC.

Materials and Methods A total of 652 patients from multi-center were enrolled and randomly allocated into primary (center A, n=375), internal (center A, n=161) and external validation (center B and C, n=116) cohorts. The baseline data were extracted from the medical records. The tumor diameter, LNM, disruption of the cervical stromal ring (DCSR) and parametrial invasion (PMI) were measured and assessed on MRI. The regions of interest (ROIs) were manually delineated along the tumor margin on the slice of T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and CE-T1WI with the largest tumor based on ITK-SNAP. The radiomics features were extracted from T2WI, DWI, and CE-T1WI. The DL features from T2WI, DWI and CE-T1WI were exported from Resnet 34 which was pretrained by 14 million natural images of ImageNet dataset. The radscore (RS) and DL score (DLS) were independently obtained after repeatability test, Pearson correlation coefficient (PCC), minimum redundancy maximum relevance (MRMR) and least absolute shrinkage and selection operator (LASSO) algorithm performed on the radiomics and DL feature sets. The DLRN was then developed by integrating the RS, DLS and independent clinicopathological factors and validated in the internal and external validation cohorts for evaluating the LNM in cervical AC/ASC.

Results For evaluating LNM, the area under the curves (AUCs) of RS and DLS were 0.75 (95% CI, 0.70-0.79) and 0.70 (95% CI, 0.65-0.75) in the primary cohort; were 0.84 (95% CI, 0.77-0.89) and 0.81 (95% CI, 0.74-0.96) in the internal validation cohort and were 0.79 (95% CI, 0.70-0.86) and 0.78 (95% CI, 0.69-0.85) in the external validation cohort, respectively. The DLRN was constructed by integrating RS, DLS, menopause and FIGO stage. The AUC, sensitivity (SEN), and specificity (SPE) of DLRN were 0.79 (95% CI, 0.74-0.83), 67.9%, and 78.4% in the primary cohort; 0.87 (95% CI, 0.81-0.92), 92.0% and 71.2% in the internal validation cohort; 0.86 (95% CI, 0.78-0.91), 89.3% and 71.6% in the external validation cohort, respectively. The AUC of DLRN was significantly higher than those of RS, DLS and clinical model in the primary and internal validation cohorts (all P < 0.05); than that of clinical model in the external validation cohort (P = 0.011).

Conclusions The nomogram of DLRN achieved good diagnostic performance than RS, DLS and clinical model. It can accurately evaluate LNM and further facilitate clinical decision-making on radiotherapy or RH, and prognosis evaluation in cervical AC/ASC.

Image/Figure:

Click to view full size

Image/Figure Caption:

The construction workflow of DLRN. (DLRN, deep learning radiomics nomogram)

Author

Meiling Xiao
Zhejiang University
0% Complete