Dataset Information

Clinical value of radiomics and machine learning in breast ultrasound: a multicenter study for differential diagnosis of benign and malignant lesions.

ABSTRACT:

Objectives

We aimed to assess the performance of radiomics and machine learning (ML) for classification of non-cystic benign and malignant breast lesions on ultrasound images, compare ML's accuracy with that of a breast radiologist, and verify if the radiologist's performance is improved by using ML.

Methods

Our retrospective study included patients from two institutions. A total of 135 lesions from Institution 1 were used to train and test the ML model with cross-validation. Radiomic features were extracted from manually annotated images and underwent a multistep feature selection process. Not reproducible, low variance, and highly intercorrelated features were removed from the dataset. Then, 66 lesions from Institution 2 were used as an external test set for ML and to assess the performance of a radiologist without and with the aid of ML, using McNemar's test.

Results

After feature selection, 10 of the 520 features extracted were employed to train a random forest algorithm. Its accuracy in the training set was 82% (standard deviation, SD, ± 6%), with an AUC of 0.90 (SD ± 0.06), while the performance on the test set was 82% (95% confidence intervals (CI) = 70-90%) with an AUC of 0.82 (95% CI = 0.70-0.93). It resulted in being significantly better than the baseline reference (p = 0.0098), but not different from the radiologist (79.4%, p = 0.815). The radiologist's performance improved when using ML (80.2%), but not significantly (p = 0.508).

Conclusions

A radiomic analysis combined with ML showed promising results to differentiate benign from malignant breast lesions on ultrasound images.

Key points

• Machine learning showed good accuracy in discriminating benign from malignant breast lesions • The machine learning classifier's performance was comparable to that of a breast radiologist • The radiologist's accuracy improved with machine learning, but not significantly.

SUBMITTER: Romeo V

PROVIDER: S-EPMC8589755 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:The aim of the present study was to assess the performance of contrast-enhanced ultrasound in distinguishing between malignant and benign breast lesions and the diagnostic value of its clinical application. A total of 52 cases with malignant breast tumors and 73 cases with benign breast lesions were included in the study. Time-intensity curves (TICs) for contrast-enhanced ultrasound were recorded, and the perfusion parameters were obtained and analyzed. Typical features of malignant breast tumors included irregular shape and vascular morphology, uneven contrast agent distribution, filling defects and contrast agent retention, 'fast-out' wash-out mode, unclear boundaries and uneven internal echo. Benign lesions were characterized by 'slow-out' or synchronous wash-out mode. Regarding perfusion, the starting time of the perfusion of the Sone-Vue microbubble contrast (always 20-30 sec) and time to peak (TTP) were significantly earlier for the malignant lesions, while the wash-out time was later. A significantly greater peak intensity, rising slope and area under the TIC were observed for the malignant breast lesions. All of the malignant breast lesions exhibited an enlarged focus scope on ultrasound, while no obvious focus scope enhancement was observed for benign breast lesions. Furthermore, the TICs of 88.4% of malignant breast lesions were of the fast-rising and slow-declining type, while the TICs of 75.3 and 17.8% of the benign breast lesions were of the slow-rising and fast-declining, and fast-rising and fast-declining type, respectively. Receiver operating characteristics analysis indicated that the TTP, wash-out time and rising slope might contribute to the differential diagnosis between malignant and benign breast lesions. In conclusion, TIC parameters of contrast-enhanced ultrasound have promising clinical value in differentiating between malignant and benign breast lesions. The TTP, wash-out time and rising slope may contribute to the diagnosis of patients with breast lesions to facilitate timely treatment and prognostication of breast cancer patients.

Project description:Background The classification of Breast Imaging Reporting and Data System 4A (BI-RADS 4A) lesions is mostly based on the personal experience of doctors and lacks specific and clear classification standards. The development of artificial intelligence (AI) provides a new method for BI-RADS categorisation. We analysed the ultrasonic morphological and texture characteristics of BI-RADS 4A benign and malignant lesions using AI, and these ultrasonic characteristics of BI-RADS 4A benign and malignant lesions were compared to examine the value of AI in the differential diagnosis of BI-RADS 4A benign and malignant lesions. Methods A total of 206 lesions of BI-RADS 4A examined using ultrasonography were analysed retrospectively, including 174 benign lesions and 32 malignant lesions. All of the lesions were contoured manually, and the ultrasonic morphological and texture features of the lesions, such as circularity, height-to-width ratio, margin spicules, margin coarseness, margin indistinctness, margin lobulation, energy, entropy, grey mean, internal calcification and angle between the long axis of the lesion and skin, were calculated using grey level gradient co-occurrence matrix analysis. Differences between benign and malignant lesions of BI-RADS 4A were analysed. Results Significant differences in margin lobulation, entropy, internal calcification and ALS were noted between the benign group and malignant group (P?=?0.013, 0.045, 0.045, and 0.002, respectively). The malignant group had more margin lobulations and lower entropy compared with the benign group, and the benign group had more internal calcifications and a greater angle between the long axis of the lesion and skin compared with the malignant group. No significant differences in circularity, height-to-width ratio, margin spicules, margin coarseness, margin indistinctness, energy, and grey mean were noted between benign and malignant lesions. Conclusions Compared with the naked eye, AI can reveal more subtle differences between benign and malignant BI-RADS 4A lesions. These results remind us carefully observation of the margin and the internal echo is of great significance. With the help of morphological and texture information provided by AI, doctors can make a more accurate judgment on such atypical benign and malignant lesions.

Project description:ObjectiveTo develop and prospective validate an ultrasound (US) prediction model to differentiate between benign and malignant subpleural pulmonary lesions (SPLs).MethodsThis study was conducted retrospectively from July 2017 to December 2018 (development cohort [DC], n = 592) and prospectively from January to April 2019 (validation cohort [VC], n = 220). A total of 18 parameters of B-mode US and contrast-enhanced US (CEUS) were acquired. Based on the DC, a model was developed using binary logistic regression. Then its discrimination and calibration were verified internally in the DC and externally in the VC, and its diagnostic performance was compared with those of the existing US diagnostic criteria in the two cohorts. The reference criteria were from the comprehensive diagnosis of clinical-radiological-pathological made by two senior respiratory physicians.ResultsThe model was eventually constructed with 6 parameters: the angle between lesion border and thoracic wall, basic intensity, lung-lesion arrival time difference, ratio of arrival time difference, vascular sign, and non-enhancing region type. In both internal and external validation, the model provided excellent discrimination of benign and malignant SPLs (C-statistic: 0.974 and 0.980 respectively), which is higher than that of "lesion-lung AT difference ≥ 2.5 s" (C-statistic: 0.842 and 0.777 respectively, P <0.001) and "AT ≥ 10 s" (C-statistic: 0.688 and 0.641 respectively, P <0.001) and the calibration curves of the model showed good agreement between actual and predictive malignancy probabilities. As for the diagnosis performance, the sensitivity and specificity of the model [sensitivity: 94.82% (DC) and 92.86% (VC); specificity: 92.42% (DC) and 92.59% (VC)] were higher than those of "lesion-lung AT difference ≥ 2.5 s" [sensitivity: 88.11% (DC) and 80.36% (VC); specificity: 80.30% (DC) and 75.00% (VC)] and "AT ≥ 10 s" [sensitivity: 64.94% (DC) and 61.61% (VC); specificity: 72.73% (DC) and 66.67% (VC)].ConclusionThe prediction model integrating multiple parameters of B-mode US and CEUS can accurately predict the malignancy probability, so as to effectively differentiate between benign and malignant SPLs, and has better diagnostic performance than the existing US diagnostic criteria.Clinical trial registrationwww.chictr.org.cn, identifier ChiCTR1800019828.

Project description:BackgroundComputer-aided methods have been widely applied to diagnose lesions detected on breast MRI, but fully-automatic diagnosis using deep learning is rarely reported.PurposeTo evaluate the diagnostic accuracy of mass lesions using region of interest (ROI)-based, radiomics and deep-learning methods, by taking peritumor tissues into consideration.Study typeRetrospective.PopulationIn all, 133 patients with histologically confirmed 91 malignant and 62 benign mass lesions for training (74 patients with 48 malignant and 26 benign lesions for testing).Field strength/sequence3T, using the volume imaging for breast assessment (VIBRANT) dynamic contrast-enhanced (DCE) sequence.Assessment3D tumor segmentation was done automatically by using fuzzy-C-means algorithm with connected-component labeling. A total of 99 texture and histogram parameters were calculated for each case, and 15 were selected using random forest to build a radiomics model. Deep learning was implemented using ResNet50, evaluated with 10-fold crossvalidation. The tumor alone, smallest bounding box, and 1.2, 1.5, 2.0 times enlarged boxes were used as inputs.Statistical testsThe malignancy probability was calculated using each model, and the threshold of 0.5 was used to make a diagnosis.ResultsIn the training dataset, the diagnostic accuracy was 76% using three ROI-based parameters, 84% using the radiomics model, and 86% using ROI + radiomics model. In deep learning using the per-slice basis, the area under the receiver operating characteristic (ROC) was comparable for tumor alone, smallest and 1.2 times box (AUC = 0.97-0.99), which were significantly higher than 1.5 and 2.0 times box (AUC = 0.86 and 0.71, respectively). For per-lesion diagnosis, the highest accuracy of 91% was achieved when using the smallest bounding box, and that decreased to 84% for tumor alone and 1.2 times box, and further to 73% for 1.5 times box and 69% for 2.0 times box. In the independent testing dataset, the per-lesion diagnostic accuracy was also the highest when using the smallest bounding box, 89%.Data conclusionDeep learning using ResNet50 achieved a high diagnostic accuracy. Using the smallest bounding box containing proximal peritumor tissue as input had higher accuracy compared to using tumor alone or larger boxes.Level of evidence3 Technical Efficacy: Stage 2.

Project description:Background Elastography is a promising way to evaluate tissue differences regarding stiffness, and the stiffness of the malignant breast lesions increased at the lesion margin. However, there is a lack of data on the value of the shear wave elastography (SWE) parameters of the surrounding tissue (shell) of different diameter on the diagnosis of benign and malignant breast lesions. Therefore, the purpose of our study was to evaluate the diagnostic performance of shell elasticity in the diagnosis of benign and malignant breast lesions using SWE. Methods Between September 2016 and June 2017, women with breast lesions underwent both conventional ultrasound (US) and SWE. Elastic values of the lesions peripheral tissue were determined according to the shell size, which was automatically drawn along the edge of the lesion using the following software guidelines: (1): 1?mm; (2): 2?mm; and (3): 3?mm. Quantitative elastographic features of the inner lesions and shell, including the elasticity mean (Emean), elasticity maximum (Emax), and elasticity minimum (Emin), were calculated using an online-available software. The receiver operating characteristic curves (ROCs) of the elastographic features was analyzed to assess the diagnostic performance, and the area under curve (AUC) of each elastographic feature was obtained. Logistic regression analysis was used to predict significant factors of malignancy, permitting the design of predictive models. Results This prospective study included 63 breast lesions of 63 women. Of the 63 lesions, 33 were malignant and 30 were benign. The diagnostic performance of Emax-3shell was the highest (AUC?=?0.76) with a sensitivity of 60.6% and a specificity of 83.3%. According to stepwise logistic regression analysis, the Emax-3shell and the Emin-3shell were significant predictors of malignancy (p?<?0.05). The AUC of the predictive equation was 0.86. Conclusions SWE features, particularly the combination of Emax-3shell and Emin-3shell can improve the diagnosis of breast lesions.