Noninvasive prediction of node-positive breast cancer response to presurgical neoadjuvant chemotherapy therapy based on machine learning of axillary lymph node ultrasound

Quantitative ultrasound radiomics analysis to evaluate lymph nodes in patients with cancer: a systematic review

This systematic review aims to evaluate the role of ultrasound (US) radiomics in assessing lymphadenopathy in patients with cancer and the ability of radiomics to predict metastatic lymph node involvement. A systematic literature search was performed in the PubMed (MEDLINE), Cochrane Central Register of Controlled Trials (CENTRAL), and EMBASE (Ovid) databases up to June 13, 2023. 42 articles were included in which the lymph node mass was assessed with a US exam, and the analysis was performed using radiomics methods. From the survey of the selected articles, experimental evidence suggests that radiomics features extracted from US images can be a useful tool for predicting and characterizing lymphadenopathy in patients with breast, head and neck, and cervical cancer. This noninvasive and effective method allows the extraction of important information beyond mere morphological characteristics, extracting features that may be related to lymph node involvement. Future studies are needed to investigate the role of US-radiomics in other types of cancers, such as melanoma.

Development and validation of a combined ultrasound-pathology model to predict axillary status after neoadjuvant systemic therapy in breast cancer

Background: This study aimed to develop a combined ultrasound (US)-pathology model to predict the axillary status more accurately after NST in breast cancer. Methods: This retrospective study included breast cancer patients who received NST at the First Affiliated Hospital of Nanjing Medical University from 2015 to 2022. Clinical, US, and pathological data were collected. Univariate and multivariate logistic regression analyses were performed to identify independent predictors of axillary pathological complete response (pCR). The model was developed using the predictors and validated. Results: A total of 657 patients were enrolled in this study. Two multivariate logistic analyses were performed before and after the operation. The results showed that the clinical lymph nodes, ER status, HER2 status, chemotherapy response of primary tumor, hilum structure of axillary lymph nodes (ALNs) after NST, blood flow of ALNs after NST, vascular invasion, pathological size, and Miller-Payne grade (all p < 0.05) were independent predictors of axillary pCR. The US-based and combined US-pathology models were developed based on preoperative and postoperative information. The two models had an area under the receiver operating characteristic curve (AUC) of 0.821 and 0.883, respectively, which was significantly higher than that of the fine-needle aspiration model (AUC: 0.735). Conclusion: In this study, based on the US-based model, a combined model incorporating the characteristics of ALNs under US and breast pathology was developed and validated to predict axillary pCR.

Non-invasive prediction of axillary lymph node dissection exemption in breast cancer patients post-neoadjuvant therapy: A radiomics and deep learning analysis on longitudinal DCE-MRI data

PURPOSE

In breast cancer (BC) patients with clinical axillary lymph node metastasis (cN+) undergoing neoadjuvant therapy (NAT), precise axillary lymph node (ALN) assessment dictates therapeutic strategy. There is a critical demand for a precise method to assess the axillary lymph node (ALN) status in these patients.

MATERIALS AND METHODS

A retrospective analysis was conducted on 160 BC patients undergoing NAT at Fujian Medical University Union Hospital. We analyzed baseline and two-cycle reassessment dynamic contrast-enhanced MRI (DCE-MRI) images, extracting 3668 radiomic and 4096 deep learning features, and computing 1834 delta-radiomic and 2048 delta-deep learning features. Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), RandomForest, and Multilayer Perceptron (MLP) algorithms were employed to develop risk models and were evaluated using 10-fold cross-validation.

RESULTS

Of the patients, 61 (38.13 %) achieved ypN0 status post-NAT. Univariate and multivariable logistic regression analyses revealed molecular subtypes and Ki67 as pivotal predictors of achieving ypN0 post-NAT. The SVM-based "Data Amalgamation" model that integrates radiomic, deep learning features, and clinical data, exhibited an outstanding AUC of 0.986 (95 % CI: 0.954-1.000), surpassing other models.

CONCLUSION

Our study illuminates the challenges and opportunities inherent in breast cancer management post-NAT. By introducing a sophisticated, SVM-based "Data Amalgamation" model, we propose a way towards accurate, dynamic ALN assessments, offering potential for personalized therapeutic strategies in BC.

Radiomics in breast cancer: Current advances and future directions

Breast cancer is a common disease that causes great health concerns to women worldwide. During the diagnosis and treatment of breast cancer, medical imaging plays an essential role, but its interpretation relies on radiologists or clinical doctors. Radiomics can extract high-throughput quantitative imaging features from images of various modalities via traditional machine learning or deep learning methods following a series of standard processes. Hopefully, radiomic models may aid various processes in clinical practice. In this review, we summarize the current utilization of radiomics for predicting clinicopathological indices and clinical outcomes. We also focus on radio-multi-omics studies that bridge the gap between phenotypic and microscopic scale information. Acknowledging the deficiencies that currently hinder the clinical adoption of radiomic models, we discuss the underlying causes of this situation and propose future directions for advancing radiomics in breast cancer research.

The use of longitudinal CT-based radiomics and clinicopathological features predicts the pathological complete response of metastasized axillary lymph nodes in breast cancer

BACKGROUND

Accurate assessment of axillary status after neoadjuvant therapy for breast cancer patients with axillary lymph node metastasis is important for the selection of appropriate subsequent axillary treatment decisions. Our objectives were to accurately predict whether the breast cancer patients with axillary lymph node metastases could achieve axillary pathological complete response (pCR).

METHODS

We collected imaging data to extract longitudinal CT image features before and after neoadjuvant chemotherapy (NAC), analyzed the correlation between radiomics and clinicopathological features, and developed models to predict whether patients with axillary lymph node metastasis can achieve axillary pCR after NAC. The clinical utility of the models was determined via decision curve analysis (DCA). Subgroup analyses were also performed. Then, a nomogram was developed based on the model with the best predictive efficiency and clinical utility and was validated using the calibration plots.

RESULTS

A total of 549 breast cancer patients with metastasized axillary lymph nodes were enrolled in this study. 42 independent radiomics features were selected from LASSO regression to construct a logistic regression model with clinicopathological features (LR radiomics-clinical combined model). The AUC of the LR radiomics-clinical combined model prediction performance was 0.861 in the training set and 0.891 in the testing set. For the HR + /HER2 - , HER2 + , and Triple negative subtype, the LR radiomics-clinical combined model yields the best prediction AUCs of 0.756, 0.812, and 0.928 in training sets, and AUCs of 0.757, 0.777 and 0.838 in testing sets, respectively.

CONCLUSIONS

The combination of radiomics features and clinicopathological characteristics can effectively predict axillary pCR status in NAC breast cancer patients.

A real-world clinicopathological model for predicting pathological complete response to neoadjuvant chemotherapy in breast cancer

Purpose

This study aimed to develop and validate a clinicopathological model to predict pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in breast cancer patients and identify key prognostic factors.

Methods

This retrospective study analyzed data from 279 breast cancer patients who received NAC at Zhejiang Provincial People's Hospital from 2011 to 2021. Additionally, an external validation dataset, comprising 50 patients from Lanxi People's Hospital and Second Affiliated Hospital, Zhejiang University School of Medicine from 2022 to 2023 was utilized for model verification. A multivariate logistic regression model was established incorporating clinical, ultrasound features, circulating tumor cells (CTCs), and pathology variables at baseline and post-NAC. Model performance for predicting pCR was evaluated. Prognostic factors were identified using survival analysis.

Results

In the 279 patients enrolled, a pathologic complete response (pCR) rate of 27.96% (78 out of 279) was achieved. The predictive model incorporated independent predictors such as stromal tumor-infiltrating lymphocyte (sTIL) levels, Ki-67 expression, molecular subtype, and ultrasound echo features. The model demonstrated strong predictive accuracy for pCR (C-statistics/AUC 0.874), especially in human epidermal growth factor receptor 2 (HER2)-enriched (C-statistics/AUC 0.878) and triple-negative (C-statistics/AUC 0.870) subtypes, and the model performed well in external validation data set (C-statistics/AUC 0.836). Incorporating circulating tumor cell (CTC) changes post-NAC and tumor size changes further improved predictive performance (C-statistics/AUC 0.945) in the CTC detection subgroup. Key prognostic factors included tumor size >5cm, lymph node metastasis, sTIL levels, estrogen receptor (ER) status and pCR. Despite varied pCR rates, overall prognosis after standard systemic therapy was consistent across molecular subtypes.

Conclusion

The developed predictive model showcases robust performance in forecasting pCR in NAC-treated breast cancer patients, marking a step toward more personalized therapeutic strategies in breast cancer.

Fusion of transfer learning models with LSTM for detection of breast cancer using ultrasound images

Breast Cancer (BC) is one of the top reasons for fatality in women worldwide. As a result, timely identification is critical for successful therapy and excellent survival rates. Transfer Learning (TL) approaches have recently shown promise in aiding in the early recognition of BC. In this work, three TL models, MobileNetV2, ResNet50, and VGG16, were combined with LSTM to extract the features from Ultrasound Images (USIs). Furthermore, the Synthetic Minority Over-sampling Technique (SMOTE) with Tomek (SMOTETomek) was employed to balance the extracted features. The proposed method with VGG16 achieved an F1 score of 99.0 %, Matthews Correlation Coefficient (MCC) and Kappa Coefficient of 98.9 % with an Area Under Curve (AUC) of 1.0. The K-fold method was applied for cross-validation and achieved an average F1 score of 96 %. Moreover, the Gradient-weighted Class Activation Mapping (Grad-CAM) method was applied for visualization, and the Local Interpretable Model-agnostic Explanations (LIME) method was applied for interpretability. The Normal Approximation Interval (NAI) and bootstrapping methods were used to calculate Confidence Intervals (CIs). The proposed method achieved a Lower CI (LCI), Upper CI (UCI), and Mean CI (MCI) of 96.50 %, 99.75 %, and 98.13 %, respectively, with the NAI, while 95 % LCI of 93.81 %, an UCI of 96.00 %, and a bootstrap mean of 94.90 % with the bootstrap method. Furthermore, the performance of the six state-of-the-art (SOTA) TL models, such as Xception, NASNetMobile, InceptionResNetV2, MobileNetV2, ResNet50, and VGG16, were compared with the proposed method.