Added Value of Quantitative Ultrasound and Machine Learning in BI-RADS 4–5 Assessment of Solid Breast Lesions

Destrempes, François; Trop, Isabelle; Allard, Louise; Chayer, Boris; García-Duitama, Julián; Khoury, Mona El; Lalonde, Lucie; Cloutier, Guy

doi:10.1016/j.ultrasmedbio.2019.10.024

Cited by 24 publications

(23 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Studies on m-US revealed malignancy rates for BI-RADS 5 lesions ranging from 57.1% [16] to 97.5% [17] and for BI-RADS 4 lesions from 8.6% [18] to 47.8% [19]. We found malignancy rates of 93.8% in m-US and 88.2 for r-US for BI-RADS 5 lesions, which for both scanning procedures is in the upper range of malignancy rates reported for m-US.…”

Section: Discussionsupporting

confidence: 42%

Similar Diagnostic Accuracy and Reproducibility of BIRADS 4 and 5 Lesions in Radial and Meander-Like Breast Ultrasound

Brasier-Lutz¹,

Jäggi-Wickes²,

Burian³

et al. 2021

austinjwomenshealth

View full text Add to dashboard Cite

Breast ultrasound is crucial in the diagnostics of breast cancer. While meander-like Ultrasound (m-US) is most commonly used, radial Ultrasound (r- US) is emerging as suitable alternative. Breast lesions category BI-RADS 4 and 5 are suspicious and highly suggestive of malignancy, respectively, and mandate breast biopsy. We compare m-US and r-US in real-time with regard to diagnostic accuracy, examination time and the agreement in location, size and final BIRADS classification of BI-RADS 4 and 5 lesions. Out of 1948 dual examinations (m-US and r-US), 57 lesions were classified as BI-RADS 4 or 5 either by r-US or m-US or by both scanning methods. For breast lesions category BI-RADS 4 or 5, sensitivity (both scan methods 94.1%), specificity (m-US 21.7%, r-US 39.1%), cancers missed rate (both 5.9%), accuracy (m-US 64.9%, r-US 71.9%), positive predictive value (m-US 64.0%, r-US 69.6%) and negative predictive value (both 100%) were similar. In m-US, the malignancy rate for category BI-RADS 5 was 93.8% versus for 50.0% for BI-RADS 4 whereas in r-US, malignancy rates were 88.2% and 58.6% for category BI-RADS 5 and 4, respectively. The examination was significantly shorter (p<0.01) for r-US (13.6 minutes) compared to m-US (27.8 minutes). Our results support radial ultrasound as an alternative to meander-like ultrasound in breast lesions category BI-RADS 4 and 5 where patients benefit from a significantly shorter examination time.

show abstract

Section: Discussionsupporting

confidence: 42%

Similar Diagnostic Accuracy and Reproducibility of BIRADS 4 and 5 Lesions in Radial and Meander-Like Breast Ultrasound

Brasier-Lutz¹,

Jäggi-Wickes²,

Burian³

et al. 2021

austinjwomenshealth

View full text Add to dashboard Cite

show abstract

“…Tsui et al analyzed the statistics of backscattered echo envelope using Nakagami statistical model, attaining 92% sensitivity, 72% specificity, and 82% accuracy in the characterization of 100 patients with breast tumours (50 benign and 50 malignant) [73]. Furthermore, Destrempes et al explored various combinations of features from shear wave elastography (SWE), RF spectral analysis, and echo envelope statistical analysis, along with BI-RADS score in the classification of 103 suspicious solid breast lesions from 103 patients (BI-RADS 3-4) [74]. They observed that the combination of SWE, QUS, and BI-RADS scoring led to an AUC of 0.97, with 76% specificity at 98% sensitivity [74].…”

Section: Discussionmentioning

confidence: 99%

“…Furthermore, Destrempes et al explored various combinations of features from shear wave elastography (SWE), RF spectral analysis, and echo envelope statistical analysis, along with BI-RADS score in the classification of 103 suspicious solid breast lesions from 103 patients (BI-RADS 3-4) [74]. They observed that the combination of SWE, QUS, and BI-RADS scoring led to an AUC of 0.97, with 76% specificity at 98% sensitivity [74]. In addition, Dobruch-Sobczak et al also found that the combination of echo envelope statistics features and BI-RADS scoring achieved 100% sensitivity, 55% specificity, and an AUC of 0.97 in the classification of 107 solid or cystic-solid breast lesions from 78 patients [75].…”

Section: Discussionmentioning

confidence: 99%

Comparison of methods for texture analysis of QUS parametric images in the characterization of breast lesions

Osapoetra¹,

Chan²,

Tran³

et al. 2020

PLoS ONE

View full text Add to dashboard Cite

Purpose Accurate and timely diagnosis of breast carcinoma is very crucial because of its high incidence and high morbidity. Screening can improve overall prognosis by detecting the disease early. Biopsy remains as the gold standard for pathological confirmation of malignancy and tumour grading. The development of diagnostic imaging techniques as an alternative for the rapid and accurate characterization of breast masses is necessitated. Quantitative ultrasound (QUS) spectroscopy is a modality well suited for this purpose. This study was carried out to evaluate different texture analysis methods applied on QUS spectral parametric images for the characterization of breast lesions. Methods Parametric images of mid-band-fit (MBF), spectral-slope (SS), spectral-intercept (SI), average scatterer diameter (ASD), and average acoustic concentration (AAC) were determined using QUS spectroscopy from 193 patients with breast lesions. Texture methods were used to quantify heterogeneities of the parametric images. Three statistical-based approaches for texture analysis that include Gray Level Co-occurrence Matrix (GLCM), Gray Level Run-length Matrix (GRLM), and Gray Level Size Zone Matrix (GLSZM) methods were evaluated. QUS and texture-parameters were determined from both tumour core and a 5-mm tumour margin and were used in comparison to histopathological analysis in order to classify breast lesions as either benign or malignant. We developed a diagnostic model using different classification algorithms including linear discriminant analysis (LDA), k-nearest neighbours (KNN), support vector machine with radial basis function kernel (SVM-RBF), and an artificial neural network (ANN). Model performance was evaluated using leave-one-out cross-validation (LOOCV) and hold-out validation. Results Classifier performances ranged from 73% to 91% in terms of accuracy dependent on tumour margin inclusion and classifier methodology. Utilizing information from tumour core alone, the ANN achieved the best classification performance of 93% sensitivity, 88% specificity, 91% accuracy, 0.95 AUC using QUS parameters and their GLSZM texture features. Conclusions A QUS-based framework and texture analysis methods enabled classification of breast lesions with >90% accuracy. The results suggest that optimizing method for extracting discriminative textural features from QUS spectral parametric images can improve classification performance. Evaluation of the proposed technique on a larger cohort of patients with proper validation technique demonstrated the robustness and generalization of the approach.

show abstract

“…It has been reported that analysis using a random forest-trained classifier shows that up to 75.9% of the BI-RADS Class 4 and 5 lesions that were later identified as benign have the potential to avoid biopsies Capacity, while tumor screening sensitivity reached 98%. Classifiers based on the BI-RADS category combined with elastography and QUS function perform best on BI-RADS categories 4 and 5 [23] . After reading a lot of relevant literature, I found that the related research and articles combining AI and breast ultrasound are rare compared to X-ray photography.…”

Section: Ai For Breast Ultrasoundmentioning

confidence: 99%

Application of Artificial Intelligence in Breast Medical Imaging Diagnosis

2020

AJOCRR

View full text Add to dashboard Cite

Review Article AJCRR (2020) 4:12 Application of Artificial Intelligence in Breast Medical Imaging Diagnosis According to the latest report on urban cancer in China published in 2017: China is one of the countries with the fastest growing incidence of breast cancer, and the age of onset has gradually become younger. Newly diagnosed breast cancer patients in China account for 12.2% of new breast cancer patients worldwide, and the mortality rate is 9.6%. A large amount of clinical experience has proven that the survival rate of breast cancer detected at an early stage is significantly higher than that detected at an advanced stage. Imaging examination is an important method for early detection of breast cancer. With the advent of the Artificial Intelligent, the method of AI + medical imaging has been widely used in lungs, breasts, heart, skull, liver, prostate, bones and other parts. Methods used in breast cancer screening include: breast self-examination and clinical physical examination, mammography of mammography of the breast, ultrasound of the breast, and magnetic resonance imaging (MRI) of the breast. The advantages and disadvantages have been reflected in the development and application in recent years. This article will review the advantages and disadvantages of combined diagnosis of AI and breast medicine. It is hoped that the artificial intelligence of medical imaging screening for breast diseases has a brighter and broader prospect.

show abstract

Added Value of Quantitative Ultrasound and Machine Learning in BI-RADS 4–5 Assessment of Solid Breast Lesions

Cited by 24 publications

References 27 publications

Similar Diagnostic Accuracy and Reproducibility of BIRADS 4 and 5 Lesions in Radial and Meander-Like Breast Ultrasound

Similar Diagnostic Accuracy and Reproducibility of BIRADS 4 and 5 Lesions in Radial and Meander-Like Breast Ultrasound

Comparison of methods for texture analysis of QUS parametric images in the characterization of breast lesions

Application of Artificial Intelligence in Breast Medical Imaging Diagnosis

Contact Info

Product

Resources

About