Deep learning radiopathomics based on preoperative US images and biopsy whole slide images can distinguish between luminal and non-luminal tumors in early-stage breast cancers.

BACKGROUND: For patients with early-stage breast cancers, neoadjuvant treatment is recommended for non-luminal tumors instead of luminal tumors. Preoperative distinguish between luminal and non-luminal cancers at early stages will facilitate treatment decisions making. However, the molecular immunohistochemical subtypes based on biopsy specimens are not always consistent with final results based on surgical specimens due to the high intra-tumoral heterogeneity. Given that, we aimed to develop and validate a deep learning radiopathomics (DLRP) model to preoperatively distinguish between luminal and non-luminal breast cancers at early stages based on preoperative ultrasound (US) images, and hematoxylin and eosin (H&E)-stained biopsy slides. METHODS: This multicentre study included three cohorts from a prospective study conducted by our team and registered on the Chinese Clinical Trial Registry (ChiCTR1900027497). Between January 2019 and August 2021, 1809 US images and 603 H&E-stained whole slide images (WSIs) from 603 patients with early-stage breast cancers were obtained. A Resnet18 model pre-trained on ImageNet and a multi-instance learning based attention model were used to extract the features of US and WSIs, respectively. An US-guided Co-Attention module (UCA) was designed for feature fusion of US and WSIs. The DLRP model was constructed based on these three feature sets including deep learning US feature, deep learning WSIs feature and UCA-fused feature from a training cohort (1467 US images and 489 WSIs from 489 patients). The DLRP model's diagnostic performance was validated in an internal validation cohort (342 US images and 114 WSIs from 114 patients) and an external test cohort (270 US images and 90 WSIs from 90 patients). We also compared diagnostic efficacy of the DLRP model with that of deep learning radiomics model and deep learning pathomics model in the external test cohort. FINDINGS: The DLRP yielded high performance with area under the curve (AUC) values of 0.929 (95% CI 0.865-0.968) in the internal validation cohort, and 0.900 (95% CI 0.819-0.953) in the external test cohort. The DLRP also outperformed deep learning radiomics model based on US images only (AUC 0.815 [0.719-0.889], p = 0.027) and deep learning pathomics model based on WSIs only (AUC 0.802 [0.704-0.878], p = 0.013) in the external test cohort. INTERPRETATION: The DLRP can effectively distinguish between luminal and non-luminal breast cancers at early stages before surgery based on pretherapeutic US images and biopsy H&E-stained WSIs, providing a tool to facilitate treatment decision making in early-stage breast cancers. FUNDING: Natural Science Foundation of Guangdong Province (No. 2023A1515011564), and National Natural Science Foundation of China (No. 91959127; No. 81971631).

Contrast-enhanced ultrasound liver imaging reporting and data system v2017: patient outcomes after treatment for early-stage hepatocellular carcinoma nodules with category 3-5 and category M.

OBJECTIVE: To evaluate correlation between contrast-enhanced ultrasonography Liver Imaging Reporting and Data System (CEUS LI-RADS; v. 2017) categories (LR 3-5 vs LR-M) and outcomes in patients with early-stage hepatocellular carcinoma (HCC) after initial therapy. METHODS: In this retrospective study, 272 patients with high risks for HCC and solitary clinically or pathologically confirmed HCC were identified between January 2010 and December 2015. Patients were initially treated by resection and radiofrequency ablation (RFA) according to the Barcelona Clinic Liver Cancer staging system and were followed up until December 31, 2018. Recurrence-free survival (RFS) and overall survival (OS) were compared between nodules assigned as LR 3-5 or LR M according to CEUS LI-RADS v. 2017 by using the Kaplan-Meier curve, log-rank test, and Cox proportional hazard model. RESULTS: Early washout is the key determinating whether a nodule is classed as LR-M. Treatment procedures and LI-RADS category showed an independent correlation with OS and RFS (p < 0.05). LR 3-5 category were more correlated with better OS (88.6 months and 74.2 months, respectively; p = 0.017) compared with LR-M. Surgical resection demonstrated longer OS and RFS than RFA in LR-M patients and longer OS in LR 3-5 patients (p < 0.05). Besides, there was no significantly difference in OS and RFS between two categories in resection (p > 0.05), while for patients treated with RFA, LR 3-5 patients showed significant longer OS and RFS than LR-M patients (p < 0.05). CONCLUSION: Patients with HCC assigned as LR-M showed worse RFS and OS and surgical resection tended to be a more effective treatment for these patients. ADVANCES IN KNOWLEDGE: Putting forward a theory that CEUS LI-RADS categories could independently predict the outcome for patients with solitary HCC at early-stage after initial treatment.

Deep Learning of Liver Contrast-Enhanced Ultrasound to Predict Microvascular Invasion and Prognosis in Hepatocellular Carcinoma.

Background and Aims: Microvascular invasion (MVI) is a well-known risk factor for poor prognosis in hepatocellular carcinoma (HCC). This study aimed to develop a deep convolutional neural network (DCNN) model based on contrast-enhanced ultrasound (CEUS) to predict MVI, and thus to predict prognosis in patients with HCC. Methods: A total of 436 patients with surgically resected HCC who underwent preoperative CEUS were retrospectively enrolled. Patients were divided into training (n = 301), validation (n = 102), and test (n = 33) sets. A clinical model (Clinical model), a CEUS video-based DCNN model (CEUS-DCNN model), and a fusion model based on CEUS video and clinical variables (CECL-DCNN model) were built to predict MVI. Survival analysis was used to evaluate the clinical performance of the predicted MVI. Results: Compared with the Clinical model, the CEUS-DCNN model exhibited similar sensitivity, but higher specificity (71.4% vs. 38.1%, p = 0.03) in the test group. The CECL-DCNN model showed significantly higher specificity (81.0% vs. 38.1%, p = 0.005) and accuracy (78.8% vs. 51.5%, p = 0.009) than the Clinical model, with an AUC of 0.865. The Clinical predicted MVI could not significantly distinguish OS or RFS (both p > 0.05), while the CEUS-DCNN predicted MVI could only predict the earlier recurrence (hazard ratio [HR] with 95% confidence interval [CI 2.92 [1.1-7.75], p = 0.024). However, the CECL-DCNN predicted MVI was a significant prognostic factor for both OS (HR with 95% CI: 6.03 [1.7-21.39], p = 0.009) and RFS (HR with 95% CI: 3.3 [1.23-8.91], p = 0.011) in the test group. Conclusions: The proposed CECL-DCNN model based on preoperative CEUS video can serve as a noninvasive tool to predict MVI status in HCC, thereby predicting poor prognosis.

Quantitative analysis of shear wave elastic heterogeneity for prediction of lymphovascular invasion in breast cancer.

OBJECTIVE: To evaluate the correlation between elastic heterogeneity (EH) and lymphovascular invasion (LVI) in breast cancers and assess the clinical value of using EH to predict LVI pre-operatively. METHODS: This retrospective study consisted of 376 patients with breast cancers that had undergone shear wave elastography (SWE) with virtual touch tissue imaging quantification between June 2017 and June 2018. The EH was determined as the difference between the averaged three highest and three lowest shear wave value. Clinicalpathological parameters including histological type and grades, LVI, axillary lymph node status and molecular markers (estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2 and Ki-67) were reviewed and recorded. Relationship EH and clinicalpathological parameters was investigated respectively. The diagnostic performance of EH in distinguishing LVI or not was analyzed. RESULTS: At multivariate regression analysis, only EH (p = 0.017) was positively correlated with LVI in all tumors. EH (p = 0.003) and Ki-67 (p = 0.025) were positively correlated with LVI in tumors ≤ 2 cm. None of clinicalpathological parameters were correlated with LVI in tumors > 2 cm (p > 0.05 for all). Using EH to predict LVI in tumors ≤ 2 cm, the sensitivity and negative predictive value were 93 and 89% respectively. CONCLUSION: EH has the potential to be served as an imaging biomarker to predict LVI in breast cancer especially for tumors ≤ 2 cm. ADVANCES IN KNOWLEDGE: There was no association between LVI and other most commonly used elastic features such as SWVmean and SWVmax. Elastic heterogeneity is an independent predictor of LVI, so it can provide additional prognostic information for routine preoperative breast cancer assessment.For tumors ≤ 2cm, using EH value higher than 1.36 m/s to predict LVI involvement, the sensitivity and negative predictive value can reach to 93% and 89%, respectively, suggesting that breast cancer with negative EH value was more likely to be absent of LVI.

Diagnosis of Non-Hepatocellular Carcinoma Malignancies in Patients With Risks for Hepatocellular Carcinoma: CEUS LI-RADS Versus CT/MRI LI-RADS.

OBJECTIVE: Data regarding direct comparison of contrast-enhanced ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS) and Computed Tomography/Magnetic Resonance Imaging (CT/MR) LI-RADS in diagnosis of non-hepatocelluar carcinoma (non-HCC) malignancies remain limited. Our study aimed to compare the diagnostic performance of the CEUS LI-RADS version 2017 and CT/MRI LI-RADS v2018 for diagnosing non-HCC malignancies in patients with risks for HCC. MATERIALS AND METHODS: In this retrospective study, 94 liver nodules pathologically-confirmed as non-HCC malignancies in 92 patients at risks for HCC from January 2009 to December 2018 were enrolled. The imaging features and the LI-RADS categories on corresponding CEUS and CT/MRI within 1 month were retrospectively analyzed according to the ACR CEUS LI-RADS v2017 and ACR CT/MRI LI-RADS v2018 by two radiologists in consensus for each algorithm. The sensitivity of LR-M category, inter-reader agreement and inter-modality agreement was compared between these two standardized algorithms. RESULTS: Ninety-four nodules in 92 patients (mean age, 54 years ± 10 [standard deviation] with 65 men [54 years ± 11] and 27 women [54 years ± 8]), including 56 intrahepatic cholangiocarcinomas, 34 combined hepatocellular cholangiocarcinomas, two adenosquamous carcinomas of the liver, one primary hepatic neuroendocrine carcinoma and one hepatic undifferentiated sarcoma were included. On CEUS, numbers of lesions classified as LR-3, LR-4, LR-5 and LR-M were 0, 1, 10 and 83, and on CT/MRI, the corresponding numbers were 3, 0, 14 and 77. There was no significant difference in the sensitivity of LR-M between these two standardized algorithms (88.3% of CEUS vs 81.9% of CT/MRI, p = 0.210). Seventy-seven lesions (81.9%) were classified as the same LI-RADS categories by both standardized algorithms (five for LR-5 and 72 for LR-M, kappa value = 0.307). In the subgroup analysis for ICC and CHC, no significant differences were found in the sensitivity of LR-M category between these two standardized algorithms (for ICC, 94.6% of CEUS vs 89.3% of CT/MRI, p = 0.375; for CHC, 76.5% of CEUS vs 70.6% of CT/MRI, p = 0. 649). CONCLUSION: CEUS LI-RADS v2017 and CT/MRI LI-RADS v2018 showed similar value for diagnosing non-HCC primary hepatic malignancies in patients with risks.

Deep learning radiomics can predict axillary lymph node status in early-stage breast cancer.

Accurate identification of axillary lymph node (ALN) involvement in patients with early-stage breast cancer is important for determining appropriate axillary treatment options and therefore avoiding unnecessary axillary surgery and complications. Here, we report deep learning radiomics (DLR) of conventional ultrasound and shear wave elastography of breast cancer for predicting ALN status preoperatively in patients with early-stage breast cancer. Clinical parameter combined DLR yields the best diagnostic performance in predicting ALN status between disease-free axilla and any axillary metastasis with areas under the receiver operating characteristic curve (AUC) of 0.902 (95% confidence interval [CI]: 0.843, 0.961) in the test cohort. This clinical parameter combined DLR can also discriminate between low and heavy metastatic burden of axillary disease with AUC of 0.905 (95% CI: 0.814, 0.996) in the test cohort. Our study offers a noninvasive imaging biomarker to predict the metastatic extent of ALN for patients with early-stage breast cancer.

Shear-Wave Elastography of the Breast: Added Value of a Quality Map in Diagnosis and Prediction of the Biological Characteristics of Breast Cancer.

OBJECTIVE: To determine the added value of a shear-wave elastography (SWE) quality map (QM) in the diagnosis of breast lesions and in predicting the biological characteristics of invasive breast cancer. MATERIALS AND METHODS: Between January 2016 and February 2019, this study included 368 women with 368 pathologically proven breast lesions, which appeared as poor-quality regions in the QM of SWE. To measure shear-wave velocity (SWV), seven regions of interest were placed in each lesion with and without QM guidance. Under QM guidance, poor-quality areas were avoided. Diagnostic performance was calculated for mean SWV (SWVmean), max SWV (SWVmax), and standard deviation (SD) with QM guidance (SWVmean + QM, SWVmax + QM, and SD + QM, respectively) and without QM guidance (SWVmean - QM, SWVmax - QM, and SD - QM, respectively). For invasive cancers, the relationship between SWV findings and biological characteristics was investigated with and without QM guidance. RESULTS: Of the 368 women (mean age, 47 years; SD, 10.8 years) enrolled, 159 had benign breast lesions and 209 had malignant breast lesions. SWVmean + QM (3.6 ± 1.39 m/s) and SD + QM (1.02 ± 0.84) were significantly different from SWVmean - QM (3.29 ± 1.22 m/s) and SD - QM (1.46 ± 1.06), respectively (all p < 0.001). For differential diagnosis of breast lesions, the sensitivity and areas under the receiver operating characteristic curve (AUC) of SWVmean + QM (sensitivity: 89%; AUC: 0.932) were better than those of SWVmean - QM (sensitivity, 84.2%; AUC, 0.912) (all p < 0.05). There was no significant difference in sensitivity and specificity between SD + QM and SD - QM (all p = 1.000). Among the biological characteristics of invasive cancers, lymphovascular involvement, axillary lymph node metastasis, negative estrogen receptor status, negative progesterone receptor status, positive human epidermal growth factor receptor status, and aggressive molecular subtypes showed higher SWVmean + QM (all p < 0.05), while only lymphovascular involvement showed higher SWVmean - QM (p = 0.036). CONCLUSION: The use of QM in SWE might improve the diagnostic performance for breast lesions and facilitate prediction of the biological characteristics of invasive breast cancers.

Added Value of Different Types of Elastography in Evaluating Ultrasonography Detected Breast Lesions: A Compared Study With Mammography.

BACKGROUND: The purpose of this study was to compare the diagnostic performance of ultrasonography (US) and mammography in the differential diagnosis of breast lesions after adding different types of elastography to US. PATIENTS AND METHODS: This institutional review board-approved study included 316 breast lesions in 289 women between July 2016 and July 2018. All these lesions were evaluated with conventional US, elastography, and mammography before biopsy or surgery. Elastography, including elasticity imaging (EI), virtual touch tissue imaging (VTI), and virtual touch imaging quantification (VTIQ), were used to downgrade US Breast Imaging-Reporting and Data System category 4A lesions. Diagnostic performances were calculated for mammography, US elastography, and the combination of US and elastography. RESULTS: The sensitivity of US (100%) was significantly higher than that of mammography (84.6%; P < .001), but the specificity of US (14.5%) was significantly lower than that of mammography (59.1%; P < .001). After adding EI, VTI, and VTIQ to US, the specificity was significantly increased from 14.5% to 69.4%, 72.6%, and 78.0%, respectively (P < .001), and were significantly higher than that of mammography (P = .043, P = .006, and P < .001, respectively). The sensitivity of US + EI (96.2%) and US + VTI (96.2%) was lower than that of US alone, although not significantly (100%; P = .063 and P = .063, respectively). CONCLUSION: The addition of different types of elastography to US improved the diagnostic performance in the differential diagnosis of breast lesions when compared with mammography.