Ultrasound features and differential diagnosis for superficial nodular fasciitis.

BACKGROUND: Nodular fasciitis (NF) has nonspecific clinical manifestations and is often misdiagnosed as sarcoma. The investigations of imaging methods for NF were limited. OBJECTIVE: To analyze the ultrasound (US) features of NF, and to evaluate the diagnostic value of US for NF. MATERIALS AND METHODS: A total of 61 NF patients were recruited retrospectively, and 551 lesions in the subcutaneous fat layer were included for comparison. We evaluated the ultrasound features of the patients and divided the NF cases into three types. Chi-square test or Fisher exact test were conducted to detect the potential difference in the distributions of three types in the two groups. RESULTS: Among the 61 NF cases, 65.6% were in the upper extremities (n = 40). The proportion of type 1, 2, and 3 were 57.4%, 24.6%, and 18.0%, respectively. NF were significantly more likely locating in the upper extremities than the other soft tissue tumors (p < 0.001). Type 1 and type 2 of sonographic features were significantly more commonly observed in NF than other soft tissue tumors among the three types (p < 0.001). CONCLUSION: The type 1 and type 2 of US features can help to distinguish NF from other lesions. US has great potential to improve the diagnostic accuracy and reduce the unnecessary surgery.

Artificial Intelligence Medical Ultrasound Equipment: Application of Breast Lesions Detection.

Breast cancer ranks first among cancers affecting women's health. Our work aims to realize the intelligence of the medical ultrasound equipment with limited computational capability, which is used for the assistant detection of breast lesions. We embed the high-computational deep learning algorithm into the medical ultrasound equipment with limited computational capability by two techniques: (1) lightweight neural network: considering the limited computational capability of ultrasound equipment, a lightweight neural network is designed, which greatly reduces the amount of calculation. And we use the technique of knowledge distillation to train the low-precision network helped with the high-precision network; (2) asynchronous calculations: consider four frames of ultrasound images as a group; the image of the first frame of each group is used as the input of the network, and the result is respectively fused with the images of the fourth to seventh frames. An amount of computation of 30 GFLO/frame is required for the proposed lightweight neural network, about 1/6 of that of the large high-precision network. After trained from scratch using the knowledge distillation technique, the detection performance of the lightweight neural network (sensitivity = 89.25%, specificity = 96.33%, the average precision [AP] = 0.85) is close to that of the high-precision network (sensitivity = 98.3%, specificity = 88.33%, AP = 0.91). By asynchronous calculation, we achieve real-time automatic detection of 24 fps (frames per second) on the ultrasound equipment. Our work proposes a method to realize the intelligence of the low-computation-power ultrasonic equipment, and successfully achieves the real-time assistant detection of breast lesions. The significance of the study is as follows: (1) The proposed method is of practical significance in assisting doctors to detect breast lesions; (2) our method provides some practical and theoretical support for the development and engineering of intelligent equipment based on artificial intelligence algorithms.

Ultrasonic Diagnosis of Breast Nodules Using Modified Faster R-CNN.

Breast cancer has become the biggest threat to female health. Ultrasonic diagnosis of breast cancer based on artificial intelligence is basically a classification of benign and malignant tumors, which does not meet clinical demand. Besides, the current target detection method performs poorly in detecting small lesions, while it is clinically required to detect nodules below 2 mm. The objective of this study is to (a) propose a diagnostic method based on Breast Imaging Reporting and Data System (BI-RADS) and (b) increase its detectability of small lesions. We modified the framework of Faster R-CNN (Faster Region-based Convolutional Neural Network) by introducing multi-scale feature extraction and multi-resolution candidate bound extraction into the network. Then, it was trained using 852 images of BI-RADS C2, 739 images of C3, and 1662 images of malignancy (BI-RADS 4a/4b/4c/5/6). We compared our model with unmodified Faster R-CNN and YOLO v3 (You Only Look Once v3). The mean average precision (mAP) is significantly increased to 0.913, while its average detection speed is slightly declined to 4.11 FPS (frames per second). Meanwhile, its detectivity of small lesions is effectively improved. Moreover, we also tentatively applied our model on video sequences and got satisfactory results. We modified Faster R-CNN and trained it partly based on BI-RADS. Its detectability of lesions, as well as small nodules, was significantly improved. In view of wide coverage of dataset and satisfactory test results, our method can basically meet clinical needs.

Deep learning driven diagnosis of malignant soft tissue tumors based on dual-modal ultrasound images and clinical indexes.

Background: Soft tissue tumors (STTs) are benign or malignant superficial neoplasms arising from soft tissues throughout the body with versatile pathological types. Although Ultrasonography (US) is one of the most common imaging tools to diagnose malignant STTs, it still has several drawbacks in STT diagnosis that need improving. Objectives: The study aims to establish this deep learning (DL) driven Artificial intelligence (AI) system for predicting malignant STTs based on US images and clinical indexes of the patients. Methods: We retrospectively enrolled 271 malignant and 462 benign masses to build the AI system using 5-fold validation. A prospective dataset of 44 malignant masses and 101 benign masses was used to validate the accuracy of system. A multi-data fusion convolutional neural network, named ultrasound clinical soft tissue tumor net (UC-STTNet), was developed to combine gray scale and color Doppler US images and clinic features for malignant STTs diagnosis. Six radiologists (R1-R6) with three experience levels were invited for reader study. Results: The AI system achieved an area under receiver operating curve (AUC) value of 0.89 in the retrospective dataset. The diagnostic performance of the AI system was higher than that of one of the senior radiologists (AUC of AI vs R2: 0.89 vs. 0.84, p=0.022) and all of the intermediate and junior radiologists (AUC of AI vs R3, R4, R5, R6: 0.89 vs 0.75, 0.81, 0.80, 0.63; p <0.01). The AI system also achieved an AUC of 0.85 in the prospective dataset. With the assistance of the system, the diagnostic performances and inter-observer agreement of the radiologists was improved (AUC of R3, R5, R6: 0.75 to 0.83, 0.80 to 0.85, 0.63 to 0.69; p<0.01). Conclusion: The AI system could be a useful tool in diagnosing malignant STTs, and could also help radiologists improve diagnostic performance.

External Validation of Ultrasound Radiomics for Small (≤ 4 cm) Renal Mass Differentiation: A Comparison with Radiologists.

BACKGROUND: Renal cell carcinoma, especially in small renal masses (≤ 4 cm) (SRM), has increased. Pathological analysis revealed a high proportion of benign masses, highlighting the urgent need for precise SRM differentiation. OBJECTIVES: This research aimed to independently validate the performance of machine learning-based ultrasound (US) radiomics analysis in differentiating benign from malignant SRM, and to compare its performance with that of radiologists. METHODS: A total of 499 patients from two hospitals were retrospectively included in this study and divided into two cohorts. US images were used to extract radiomics features. To obtain the most robust features, inter-observer correlation coefficient, Spearman correlation coefficient, and least absolute shrinkage and selection operator methods were applied for feature selection. Three models were developed in the training data using the stochastic gradient boosting algorithm, including a clinical model, a radiomics model, and a combined model that integrated clinical factors and radiomics features. The performance of these models was evaluated in the independent external validation data, including discrimination, calibration, and clinical usefulness, and compared with pooled radiologists' assessments. RESULTS: The AUCs of the clinical, radiomics, and combined models were 0.844, 0.942, and 0.954, respectively. The radiomics and combined models significantly outperformed the clinical model (all p < 0.05), while no significant difference was observed between them (p = 0.32). The radiomics and combined models showed good discrimination and calibration. Decision curve analysis exhibited that the combined model had clinical usefulness. Compared with the pooled radiologists' assessment (AUC, 0.799), the combined model showed superior classification results (p < 0.01) and higher specificity (p < 0.01) with similar sensitivity (p = 0.62). CONCLUSION: The combined model incorporating clinical factors and radiomics features accurately distinguished benign from malignant SRM.

Benefit of Using Both Ultrasound Imaging and Clinical Information for Predicting Malignant Soft Tissue Tumors.

OBJECTIVE: Ultrasonography (US) is the primary imaging method for soft tissue tumors (STTs), the diagnostic performance of which still requires improvement. To achieve an accurate evaluation of STTs, we built the diagnostic nomogram for STTs using the clinical and US features of patients with STTs. METHODS: A total of 613 patients with 195 malignant and 418 benign STTs were retrospectively recruited. We used a blend of clinical and ultrasonic features, as well as exclusively US features, to develop two distinct diagnostic models for STTs: the clinical-US model and the US-only model, respectively. The two models were evaluated and compared by measuring their areas under the receiver operating characteristic curve (AUC), calibration, integrated discrimination improvement (IDI) and decision curve analysis. The performance of the clinical-US model was also compared with that of two radiologists. RESULTS: The clinical-US model had better diagnostic performance than the model based on US imaging features alone (AUCs of the clinical-US and US-only models: 0.95 [0.93-0.97] vs. 0.89 [0.87-0.92], p < 0.001; IDI of the two models: 0.15 ± 0.03, p < 0.001). The clinical-US model was also superior to the two radiologists in diagnosing STTs (AUCs of clinical-US model and two radiologists: 0.95 [0.93-0.97] vs. 0.79 [0.75-0.82] and 0.83 [0.80-0.85], p < 0.001). CONCLUSION: The diagnostic model based on clinical and US imaging features had high diagnostic performance in STTs, which could help identify malignant STTs for radiologists.

Incorporating Contrast-Enhanced Ultrasound into the BI-RADS Scoring System Improves Accuracy in Breast Tumor Diagnosis: A Preliminary Study in China.

The aim of the study was to develop a scoring model incorporating the Breast Imaging Reporting and Data System (BI-RADS) and the contrast-enhanced ultrasound (CEUS) scoring system to differentiate between malignant and benign breast lesions. A total of 524 solid breast masses in 490 consecutive patients were evaluated with conventional US and CEUS in this prospective study. Each lesion was scored according to BI-RADS, CEUS, and CEUS-rerated BI-RADS. The diagnostic specificity, sensitivity and accuracy of BI-RADS were 77.9%, 88.9% and 84.0%, respectively, and the area under the receiver operating characteristic curve was 0.834. The corresponding values for rerated BI-RADS were 82.1%, 96.9%, 90.3% and 0.895. The area under the receiver operating characteristic curve of BI-RADS alone was significantly smaller than that of CEUS and the rerated BI-RADS (p = 0.008 compared with CEUS, p = 0.002 compared with rerated BI-RADS). This study indicates that rerating BI-RADS with the CEUS scoring system improves its diagnostic accuracy.