Assessment of the invariance and discriminant power of morphological features under geometric transformations for breast tumor classification

Gómez-Flores, Wilfrido; Hernández-López, Juanita

doi:10.1016/j.cmpb.2019.105173

Cited by 27 publications

(9 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Roundness was calculated using the formula: 4π × surface area/perimeter². Concavity was calculated indirectly from the measured convex closure area: (convex closure area – surface area)/convex closure area ( 24 ). Thirty cases were randomly selected from the enrolled population before assignment, and the repeatability of feature extraction was assessed using intra-observer and inter-observer intraclass correlation coefficients (ICCs).…”

Section: Methodsmentioning

confidence: 99%

A Novel Combined Nomogram Model for Predicting the Pathological Complete Response to Neoadjuvant Chemotherapy in Invasive Breast Carcinoma of No Specific Type: Real-World Study

et al. 2022

View full text Add to dashboard Cite

ObjectiveTo explore the value of a predictive model combining the multiparametric magnetic resonance imaging (mpMRI) radiomics score (RAD-score), clinicopathologic features, and morphologic features for the pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in invasive breast carcinoma of no specific type (IBC-NST).MethodsWe enrolled, retrospectively and consecutively, 206 women with IBC-NST who underwent surgery after NAC and obtained pathological results from August 2018 to October 2021. Four RAD-scores were constructed for predicting the pCR based on fat-suppression T2-weighted imaging (FS-T2WI), diffusion-weighted imaging (DWI), contrast-enhanced T1-weighted imaging (T1WI+C) and their combination, which was called mpMRI. The best RAD-score was combined with clinicopathologic and morphologic features to establish a nomogram model through binary logistic regression. The predictive performance of the nomogram was evaluated using the area under receiver operator characteristic (ROC) curve (AUC) and calibration curve. The clinical net benefit of the model was evaluated using decision curve analysis (DCA).ResultsThe mpMRI RAD-score had the highest diagnostic performance, with AUC of 0.848 among the four RAD-scores. T stage, human epidermal growth factor receptor-2 (HER2) status, RAD-score, and roundness were independent factors for predicting the pCR (P < 0.05 for all). The combined nomogram model based on these factors achieved AUCs of 0.930 and 0.895 in the training cohort and validation cohort, respectively, higher than other models (P < 0.05 for all). The calibration curve showed that the predicted probabilities of the nomogram were in good agreement with the actual probabilities, and DCA indicated that it provided more net benefit than the treat-none or treat-all scheme by decision curve analysis in both training and validation datasets.ConclusionThe combined nomogram model based on the mpMRI RAD-score combined with clinicopathologic and morphologic features may improve the predictive performance for the pCR of NAC in patients with IBC-NST.

show abstract

Section: Methodsmentioning

confidence: 99%

A Novel Combined Nomogram Model for Predicting the Pathological Complete Response to Neoadjuvant Chemotherapy in Invasive Breast Carcinoma of No Specific Type: Real-World Study

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Gómez-Flores and Hernández-López [50] proposed a CAD which helped radiologists to classify breast cancer. The proposed CAD is based on 39 morphological features which describe breast tumor shapes to distinguish whether they are benign or malignant tumors.…”

Section: Related Workmentioning

confidence: 99%

An Optimized Framework for Breast Cancer Classification Using Machine Learning

Michael

et al. 2022

BioMed Research International

View full text Add to dashboard Cite

Breast cancer, if diagnosed and treated early, has a better chance of surviving. Many studies have shown that a larger number of ultrasound images are generated every day, and the number of radiologists able to analyze this medical data is very limited. This often results in misclassification of breast lesions, resulting in a high false-positive rate. In this article, we propose a computer-aided diagnosis (CAD) system that can automatically generate an optimized algorithm. To train machine learning, we employ 13 features out of 185 available. Five machine learning classifiers were used to classify malignant versus benign tumors. The experimental results revealed Bayesian optimization with a tree-structured Parzen estimator based on a machine learning classifier for 10-fold cross-validation. The LightGBM classifier performs better than the other four classifiers, achieving 99.86% accuracy, 100.0% precision, 99.60% recall, and 99.80% for the FI score.

show abstract

“…The first classifier was based on shape parameters. In this case, the following 15 morphological features were determined based on the mass boundary outlines: depth-to-width ratio, mass area, circularity, roundness, normalized residual value, overlap ratio, convexity, orientation, long axis to short axis ratio, elliptic normalized skeleton, elliptic normalized circumference, mean of normalized radial length (NRL), standard deviation of NRL, area ratio and contour roughness [3], [10], [11]. The second classifier was trained with the texture features calculated using the gray-level cooccurrence matrix (GLCM) technique.…”

Section: B Standard Classifiersmentioning

confidence: 99%

Deep meta-learning for the selection of accurate ultrasound based breast mass classifier

Byra

Karwat

Ryzhankow

et al. 2022

2022 IEEE International Ultrasonics Symposium (IUS)

View full text Add to dashboard Cite

show abstract

Assessment of the invariance and discriminant power of morphological features under geometric transformations for breast tumor classification

Cited by 27 publications

References 37 publications

A Novel Combined Nomogram Model for Predicting the Pathological Complete Response to Neoadjuvant Chemotherapy in Invasive Breast Carcinoma of No Specific Type: Real-World Study

A Novel Combined Nomogram Model for Predicting the Pathological Complete Response to Neoadjuvant Chemotherapy in Invasive Breast Carcinoma of No Specific Type: Real-World Study

An Optimized Framework for Breast Cancer Classification Using Machine Learning

Deep meta-learning for the selection of accurate ultrasound based breast mass classifier

Contact Info

Product

Resources

About