Junior Radiologists Research Articles

Prospective study of AI-assisted prediction of breast malignancies in physical health examinations: role of off-the-shelf AI software and comparison to radiologist performance.

In physical health examinations, breast sonography is a commonly used imaging method, but it can lead to repeated exams and unnecessary biopsy due to discrepancies among radiologists and health centers. This study explores the role of off-the-shelf artificial intelligence (AI) software in assisting radiologists to classify incidentally found breast masses in two health centers. Female patients undergoing breast ultrasound examinations with incidentally discovered breast masses were categorized according to the 5th edition of the Breast Imaging Reporting and Data System (BI-RADS), with categories 3 to 5 included in this study. The examinations were conducted at two municipal health centers from May 2021 to May 2023.The final pathological results from surgical resection or biopsy served as the gold standard for comparison. Ultrasonographic images were obtained in longitudinal and transverse sections, and two junior radiologists and one senior radiologist independently assessed the images without knowing the pathological findings. The BI-RADS classification was adjusted following AI assistance, and diagnostic performance was compared using receiver operating characteristic curves. A total of 196 patients with 202 breast masses were included in the study, with pathological results confirming 107 benign and 95 malignant masses. The receiver operating characteristic curve showed that experienced breast radiologists had higher diagnostic performance in BI-RADS classification than junior radiologists, similar to AI classification (AUC = 0.936, 0.806, 0.896, and 0.950, p < 0.05). The AI software improved the accuracy, sensitivity, and negative predictive value of the adjusted BI-RADS classification for the junior radiologists' group (p< 0.05), while no difference was observed in the senior radiologist group. Furthermore, AI increased the negative predictive value for BI-RADS 4a masses and the positive predictive value for 4b masses among radiologists (p < 0.05). AI enhances the sensitivity of invasive breast cancer detection more effectively than ductal carcinoma in situ and rare subtypes of breast cancer. The AI software enhances diagnostic efficiency for breast masses, reducing the performance gap between junior and senior radiologists, particularly for BI-RADS 4a and 4b masses. This improvement reduces unnecessary repeat examinations and biopsies, optimizing medical resource utilization and enhancing overall diagnostic effectiveness.

Abstract PO1-07-08: Value of high resolution full field optical coherence tomography and dynamic cell imaging (D-FFOCT) for one-stop rapid diagnosis breast clinic

Abstract BACKGROUND: Full field optical coherence tomography combined to dynamic cell imaging (D-FFOCT) is a new, simple to use, nondestructive, quick technique than can provide sufficient spatial resolution to mimic histopathological analysis. The objective of this study was to evaluate diagnostic performance of D-FFOCT for one-stop rapid diagnosis breast clinic. METHODS: D-FFOCT was applied to fresh untreated breast and nodes biopsies. Four different readers (senior and junior radiologist, surgeon and pathologist) analyzed the samples without knowing final histological diagnosis or ACR classification. The results were compared to conventional processing and staining (Hematoxylin-eosin). RESULTS: A total of 217 biopsies were performed on 152 patients. There were 144 breast biopsies and 61 lymph nodes with 101 infiltrative cancers (49,27%), 99 benign lesions (48,29%), 3 ductal in situ carcinoma (1,46%) and 2 atypias (0,98%). The diagnostic performances results were as follow: sensitivity: 77% [0.7;0.82], specificity: 64% [0.58;0.71], PPV: 74% [0.68;0.78] and NPV: 75% [0.72;0.78]. A large images atlas was created as well as a diagnosis algorithm from the readers experience. CONCLUSION: D-FFOCT provides interesting results defining malignancy on fresh breast and nodes samples. By training with the diagnosis algorithm and the images atlas or by using machine learning, radiologists could obtain even better outcomes allowing quick detection of breast cancer and lymph node involvement. diagnosis algorithm Figure 1 Images in DCI and FFOCT Two infiltrating cancer diagnosed on pathology (A,B). Focusing on cells (C). Normal galactophoric duct (D). Adenofibroma (E) Citation Format: Alexis Simon, Yasmina Badachi, Jacques Ropers, Isaura Laurent, Lida Dong, Elisabeth Da Maia, Agnes Bourcier, Geoffroy Canlorbe, Catherine Uzan. Value of high resolution full field optical coherence tomography and dynamic cell imaging (D-FFOCT) for one-stop rapid diagnosis breast clinic [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO1-07-08.

Deep learning-assisted diagnosis of benign and malignant parotid tumors based on ultrasound: a retrospective study

BackgroundTo develop a deep learning(DL) model utilizing ultrasound images, and evaluate its efficacy in distinguishing between benign and malignant parotid tumors (PTs), as well as its practicality in assisting clinicians with accurate diagnosis.MethodsA total of 2211 ultrasound images of 980 pathologically confirmed PTs (Training set: n = 721; Validation set: n = 82; Internal-test set: n = 89; External-test set: n = 88) from 907 patients were retrospectively included in this study. The optimal model was selected and the diagnostic performance evaluation is conducted by utilizing the area under curve (AUC) of the receiver-operating characteristic(ROC) based on five different DL networks constructed at varying depths. Furthermore, a comparison of different seniority radiologists was made in the presence of the optimal auxiliary diagnosis model. Additionally, the diagnostic confusion matrix of the optimal model was calculated, and an analysis and summary of misjudged cases’ characteristics were conducted.ResultsThe Resnet18 demonstrated superior diagnostic performance, with an AUC value of 0.947, accuracy of 88.5%, sensitivity of 78.2%, and specificity of 92.7% in internal-test set, and with an AUC value of 0.925, accuracy of 89.8%, sensitivity of 83.3%, and specificity of 90.6% in external-test set. The PTs were subjectively assessed twice by six radiologists, both with and without the assisted of the model. With the assisted of the model, both junior and senior radiologists demonstrated enhanced diagnostic performance. In the internal-test set, there was an increase in AUC values by 0.062 and 0.082 for junior radiologists respectively, while senior radiologists experienced an improvement of 0.066 and 0.106 in their respective AUC values.ConclusionsThe DL model based on ultrasound images demonstrates exceptional capability in distinguishing between benign and malignant PTs, thereby assisting radiologists of varying expertise levels to achieve heightened diagnostic performance, and serve as a noninvasive imaging adjunct diagnostic method for clinical purposes.

Applying machine-learning models to differentiate benign and malignant thyroid nodules classified as C-TIRADS 4 based on 2D-ultrasound combined with five contrast-enhanced ultrasound key frames.

To apply machine learning to extract radiomics features from thyroid two-dimensional ultrasound (2D-US) combined with contrast-enhanced ultrasound (CEUS) images to classify and predict benign and malignant thyroid nodules, classified according to the Chinese version of the thyroid imaging reporting and data system (C-TIRADS) as category 4. This retrospective study included 313 pathologically diagnosed thyroid nodules (203 malignant and 110 benign). Two 2D-US images and five CEUS key frames ("2nd second after the arrival time" frame, "time to peak" frame, "2nd second after peak" frame, "first-flash" frame, and "second-flash" frame) were selected to manually label the region of interest using the "Labelme" tool. A total of 7 images of each nodule and their annotates were imported into the Darwin Research Platform for radiomics analysis. The datasets were randomly split into training and test cohorts in a 9:1 ratio. Six classifiers, namely, support vector machine, logistic regression, decision tree, random forest (RF), gradient boosting decision tree and extreme gradient boosting, were used to construct and test the models. Performance was evaluated using a receiver operating characteristic curve analysis. The area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy (ACC), and F1-score were calculated. One junior radiologist and one senior radiologist reviewed the 2D-US image and CEUS videos of each nodule and made a diagnosis. We then compared their AUC and ACC with those of our best model. The AUC of the diagnosis of US, CEUS and US combined CEUS by junior radiologist and senior radiologist were 0.755, 0.750, 0.784, 0.800, 0.873, 0.890, respectively. The RF classifier performed better than the other five, with an AUC of 1 for the training cohort and 0.94 (95% confidence interval 0.88-1) for the test cohort. The sensitivity, specificity, accuracy, PPV, NPV, and F1-score of the RF model in the test cohort were 0.82, 0.93, 0.90, 0.85, 0.92, and 0.84, respectively. The RF model with 2D-US combined with CEUS key frames achieved equivalent performance as the senior radiologist (AUC: 0.94 vs. 0.92, P = 0.798; ACC: 0.90 vs. 0.92) and outperformed the junior radiologist (AUC: 0.94 vs. 0.80, P = 0.039, ACC: 0.90 vs. 0.81) in the test cohort. Our model, based on 2D-US and CEUS key frames radiomics features, had good diagnostic efficacy for thyroid nodules, which are classified as C-TIRADS 4. It shows promising potential in assisting less experienced junior radiologists.

Fully-Automatic Detection and Diagnosis System for Thyroid Nodules Based on Ultrasound Video Sequences by Artificial Intelligence.

Interpretation of ultrasound findings of thyroid nodules is subjective and labor-intensive for radiologists. Artificial intelligence (AI) is a relatively objective and efficient technology. We aimed to establish a fully automatic detection and diagnosis system for thyroid nodules based on AI technology by analyzing ultrasound video sequences. We prospectively acquired dynamic ultrasound videos of 1067 thyroid nodules (804 for training and 263 for validation) from December 2018 to January 2021. All the patients underwent hemithyroidectomy or total thyroidectomy. Dynamic ultrasound videos were used to develop an AI system consisting of two deep learning models that could automatically detect and diagnose thyroid nodules. Average precision (AP) was used to estimate the performance of the detection model. The area under the receiver operating characteristic curve (AUC) was used to measure the performance of the diagnostic model. Location and shape were accurately detected with a high AP of 0.914 in the validation cohort. The AUC of the diagnostic model was 0.953 in the validation cohort. The sensitivity and specificity of junior and senior radiologists were 76.9% vs 78.3% and 68.4% vs 81.1%, respectively. The diagnostic performance of the AI diagnostic model was superior to that of junior radiologists (P = 0.016) and was not significantly different from that of senior radiologists (P = 0.281). We established a fully automatic detection and diagnosis system for thyroid nodules based on ultrasound video using an AI approach that can be conveniently applied to optimize the management of patients with thyroid nodules.

Interoperator reproducibility of quantitative ultrasound analysis of hepatic steatosis in participants with suspected MASLD: A prospective study

ObjectivesTo evaluate the reproducibility of tissue attenuation imaging (TAI) and tissue scatter distribution imaging (TSI) measurements in adults with suspected metabolic dysfunction-associated steatotic liver disease (MASLD) between radiologists with varying experience. Materials and methodsParticipants with suspected MASLD were prospectively recruited. TAI and TSI were performed for each participant by two radiologists with different levels of experience. Interoperability reliability was assessed on the basis of Bland-Altman analysis and intraclass correlation coefficients (ICCs). The study determined and compared the diagnostic performance of TAI and TSI with clinical prediction models using proton magnetic resonance spectroscopy (1H-MRS) as a reference. ResultsA total of 180 participants (women, n = 56; men, n = 124, mean age, 46.98 ± 14.92 years; mean BMI, 25.81 ± 4.47) were enrolled from August 2022 to September 2022. Bland-Altman plots showed only slight deviation in the TAI and TSI results of the two radiologists; there was good interoperator reproducibility for TAI (ICC = 0.92) and TSI (ICC = 0.86). Senior and junior radiologists performed examinations labeled as TAI-1 and TSI-1, and TAI-2 and TSI-2, respectively. The areas under the curves (AUCs) of TAI-1, TAI-2, TSI-1, and TAI-2 for the detection of ≥5 % hepatic steatosis were 0.90, 0.96, 0.91 and 0.96, respectively. According to ROC analysis, the diagnostic performance of both radiologists for TAI and TSI was statistically similar and superior to that of the clinical prediction model. ConclusionsTAI and TSI have good reproducibility between radiologists with different levels of experience. Meanwhile, both TAI and TSI demonstrated good diagnostic performance for hepatic steatosis (≥5%), surpassing that of clinical prediction models.

Automatic detection of thyroid nodules with a real-time artificial intelligence system in a real clinical scenario and the associated influencing factors.

At present, most articles mainly focused on the diagnosis of thyroid nodules by using artificial intelligence (AI), and there was little research on the detection performance of AI in thyroid nodules. To explore the value of a real-time AI based on computer-aided diagnosis system in the detection of thyroid nodules and to analyze the factors influencing the detection accuracy. From June 1, 2022 to December 31, 2023, 224 consecutive patients with 587 thyroid nodules were prospective collected. Based on the detection results determined by two experienced radiologists (both with more than 15 years experience in thyroid diagnosis), the detection ability of thyroid nodules of radiologists with different experience levels (junior radiologist with 1 year experience and senior radiologist with 5 years experience in thyroid diagnosis) and real-time AI were compared. According to the logistic regression analysis, the factors influencing the real-time AI detection of thyroid nodules were analyzed. The detection rate of thyroid nodules by real-time AI was significantly higher than that of junior radiologist (P = 0.013), but lower than that of senior radiologist (P = 0.001). Multivariate logistic regression analysis showed that nodules size, superior pole, outside (near carotid artery), close to vessel, echogenicity (isoechoic, hyperechoic, mixed-echoic), morphology (not very regular, irregular), margin (unclear), ACR TI-RADS category 4 and 5 were significant independent influencing factors (all P < 0.05). With the combination of real-time AI and radiologists, junior and senior radiologist increased the detection rate to 97.4% (P < 0.001) and 99.1% (P = 0.015) respectively. The real-time AI has good performance in thyroid nodule detection and can be a good auxiliary tool in the clinical work of radiologists.

Ultrasound radiomics-based artificial intelligence model to assist in the differential diagnosis of ovarian endometrioma and ovarian dermoid cyst.

Accurately differentiating between ovarian endometrioma and ovarian dermoid cyst is of clinical significance. However, the ultrasound appearance of these two diseases is variable, occasionally causing confusion and overlap with each other. This study aimed to develop a diagnostic classification model based on ultrasound radiomics to intelligently distinguish and diagnose the two diseases. We collected ovarian ultrasound images from participants diagnosed as patients with ovarian endometrioma or ovarian dermoid cyst. Feature extraction and selection were performed using the Mann-Whitney U-test, Spearman correlation analysis, and the least absolute shrinkage and selection operator (LASSO) regression. We then input the final features into the machine learning classifiers for model construction. A nomogram was established by combining the radiomic signature and clinical signature. A total of 407 participants with 407 lesions were included and categorized into the ovarian endometriomas group (n = 200) and the dermoid cyst group (n = 207). In the test cohort, Logistic Regression (LR) achieved the highest area under curve (AUC) value (0.981, 95% CI: 0.963-1.000), the highest accuracy (94.8%), and the highest sensitivity (95.5%), while LightGBM achieved the highest specificity (97.1%). A nomogram incorporating both clinical features and radiomic features achieved the highest level of performance (AUC: 0.987, 95% CI: 0.967-1.000, accuracy: 95.1%, sensitivity: 88.0%, specificity: 100.0%, PPV: 100.0%, NPV: 88.0%, precision: 93.6%). No statistical difference in diagnostic performance was observed between the radiomic model and the nomogram (P > 0.05). The diagnostic indexes of radiomic model were comparable to that of senior radiologists and superior to that of junior radiologist. The diagnostic performance of junior radiologists significantly improved with the assistance of the model. This ultrasound radiomics-based model demonstrated superior diagnostic performance compared to those of junior radiologists and comparable diagnostic performance to those of senior radiologists, and it has the potential to enhance the diagnostic performance of junior radiologists.

Prediction of clinically significant prostate cancer using radiomics models in real-world clinical practice: a retrospective multicenter study

PurposeTo develop and evaluate machine learning models based on MRI to predict clinically significant prostate cancer (csPCa) and International Society of Urological Pathology (ISUP) grade group as well as explore the potential value of radiomics models for improving the performance of radiologists for Prostate Imaging Reporting and Data System (PI-RADS) assessment.Material and methodsA total of 1616 patients from 4 tertiary care medical centers were retrospectively enrolled. PI-RADS assessments were performed by junior, senior, and expert-level radiologists. The radiomics models for predicting csPCa were built using 4 machine-learning algorithms. The PI-RADS were adjusted by the radiomics model. The relationship between the Rad-score and ISUP was evaluated by Spearman analysis.ResultsThe radiomics models made using the random forest algorithm yielded areas under the receiver operating characteristic curves (AUCs) of 0.874, 0.876, and 0.893 in an internal testing cohort and external testing cohorts, respectively. The AUC of the adjusted_PI-RADS was improved, and the specificity was improved at a slight sacrifice of sensitivity. The participant-level correlation showed that the Rad-score was positively correlated with ISUP in all testing cohorts (r > 0.600 and p < 0.0001).ConclusionsThis radiomics model resulted as a powerful, non-invasive auxiliary tool for accurately predicting prostate cancer aggressiveness. The radiomics model could reduce unnecessary biopsies and help improve the diagnostic performance of radiologists’ PI-RADS. Yet, prospective studies are still needed to validate the radiomics models further.Critical relevance statementThe radiomics model with MRI may help to accurately screen out clinically significant prostate cancer, thereby assisting physicians in making individualized treatment plans.Key points• The diagnostic performance of the radiomics model using the Random Forest algorithm is comparable to the Prostate Imaging Reporting and Data System (PI-RADS) obtained by radiologists.• The performance of the adjusted Prostate Imaging Reporting and Data System (PI-RADS) was improved, which implied that the radiomics model could be a potential radiological assessment tool.• The radiomics model lowered the percentage of equivocal cases. Moreover, the Rad-scores can be used to characterize prostate cancer aggressiveness.Graphical

Artificial intelligence assistance for fetal development: evaluation of an automated software for biometry measurements in the mid-trimester

BackgroundThis study presents CUPID, an advanced automated measurement software based on Artificial Intelligence (AI), designed to evaluate nine fetal biometric parameters in the mid-trimester. Our primary objective was to assess and compare the CUPID performance of experienced senior and junior radiologists.Materials and methodsThis prospective cross-sectional study was conducted at Shenzhen University General Hospital between September 2022 and June 2023, and focused on mid-trimester fetuses. All ultrasound images of the six standard planes, that enabled the evaluation of nine biometric measurements, were included to compare the performance of CUPID through subjective and objective assessments.ResultsThere were 642 fetuses with a mean (±SD) age of 22 ± 2.82 weeks at enrollment. In the subjective quality assessment, out of 642 images representing nine biometric measurements, 617-635 images (90.65-96.11%) of CUPID caliper placements were determined to be accurately placed and did not require any adjustments. Whereas, for the junior category, 447-691 images (69.63-92.06%) were determined to be accurately placed and did not require any adjustments. In the objective measurement indicators, across all nine biometric parameters and estimated fetal weight (EFW), the intra-class correlation coefficients (ICC) (0.843-0.990) and Pearson correlation coefficients (PCC) (0.765-0.978) between the senior radiologist and CUPID reflected good reliability compared with the ICC (0.306-0.937) and PCC (0.566-0.947) between the senior and junior radiologists. Additionally, the mean absolute error (MAE), percentage error (PE), and average error in days of gestation were lower between the senior and CUPID compared to the difference between the senior and junior radiologists. The specific differences are as follows: MAE (0.36-2.53 mm, 14.67 g) compared to (0.64- 8.13 mm, 38.05 g), PE (0.94-9.38%) compared to (1.58-16.04%), and average error in days (3.99-7.92 days) compared to (4.35-11.06 days). In the time-consuming task, CUPID only takes 0.05-0.07 s to measure nine biometric parameters, while senior and junior radiologists require 4.79-11.68 s and 4.95-13.44 s, respectively.ConclusionsCUPID has proven to be highly accurate and efficient software for automatically measuring fetal biometry, gestational age, and fetal weight, providing a precise and fast tool for assessing fetal growth and development.

A novel predict method for muscular invasion of bladder cancer based on 3D mp-MRI feature fusion

Objective. To assist urologist and radiologist in the preoperative diagnosis of non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC), we proposed a combination models strategy (CMS) utilizing multiparametric magnetic resonance imaging. Approach. The CMS includes three components: image registration, image segmentation, and multisequence feature fusion. To ensure spatial structure consistency of T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced imaging (DCE), a registration network based on patch sampling normalized mutual information was proposed to register DWI and DCE to T2WI. Moreover, to remove redundant information around the bladder, we employed a segmentation network to obtain the bladder and tumor regions from T2WI. Using the coordinate mapping from T2WI, we extracted these regions from DWI and DCE and integrated them into a three-branch dual-channel input. Finally, to fully fuse low-level and high-level features of T2WI, DWI, and DCE, we proposed a distributed multilayer fusion model for preoperative MIBC prediction with five-fold cross-validation. Main results. The study included 436 patients, of which 404 were for the internal cohort and 32 for external cohort. The MIBC was confirmed by pathological examination. In the internal cohort, the area under the curve, accuracy, sensitivity, and specificity achieved by our method were 0.928, 0.869, 0.753, and 0.929, respectively. For the urologist and radiologist, Vesical Imaging-Reporting and Data System score >3 was employed to determine MIBC. The urologist demonstrated an accuracy, sensitivity, and specificity of 0.842, 0.737, and 0.895, respectively, while the radiologist achieved 0.871, 0.803, and 0.906, respectively. In the external cohort, the accuracy of our method was 0.831, which was higher than that of the urologist (0.781) and the radiologist (0.813). Significance. Our proposed method achieved better diagnostic performance than urologist and was comparable to senior radiologist. These results indicate that CMS can effectively assist junior urologists and radiologists in diagnosing preoperative MIBC.

Application of Spinal Subtraction and Bone Background Fusion CTA in the Accurate Diagnosis and Evaluation of Spinal Vascular Malformations.

Accurate pretreatment diagnosis and assessment of spinal vascular malformations using spinal CTA are crucial for patient prognosis, but the postprocessing reconstruction may not be able to fully depict the lesions due to the complexity inherent in spinal anatomy. Our purpose was to explore the application value of the spinal subtraction and bone background fusion CTA (SSBBF-CTA) technique in precisely depicting and localizing spinal vascular malformation lesions. In this retrospective study, patients (between November 2017 and November 2022) with symptoms similar to those of spinal vascular malformations were divided into diseased (group A) and nondiseased (group B) groups. All patients underwent spinal CTA using Siemens dual-source CT. Multiplanar reconstruction; routine bone subtraction, and SSBBF-CTA images were obtained using the snygo.via and ADW4.6 postprocessing reconstruction workstations. Multiple observers researched the following 3 aspects: 1) preliminary screening capability using original images with multiplanar reconstruction CTA, 2) the accuracy and stability of the SSBBF-CTA postprocessing technique, and 3) diagnostic evaluation of spinal vascular malformations using the 3 types of postprocessing images. Diagnostic performance was analyzed using receiver operating characteristic analysis, while reader or image differences were analyzed using the Wilcoxon signed-rank test or the Kruskal-Wallis rank sum test. Forty-nine patients (groups A and B: 22 and 27 patients; mean ages, 44.0 [SD, 14.3] years and 44.6 [SD,15.2] years; 13 and 16 men) were evaluated. Junior physicians showed lower diagnostic accuracy and sensitivity using multiplanar reconstruction CTA (85.7% and 77.3%) than senior physicians (93.9% and 90.9%, 98% and 95.5%). Short-term trained juniors achieved SSBBF-CTA image accuracy similar to that of experienced physicians (P > .05). In terms of the visualization and localization of spinal vascular malformation lesions (nidus/fistula, feeding artery, and drainage vein), both multiplanar reconstruction and SSBBF-CTA outperformed routine bone subtraction CTA (P = .000). Compared with multiplanar reconstruction, SSBBF-CTA allowed less experienced physicians to achieve superior diagnostic capabilities (comparable with those of experienced radiologists) more rapidly (P < .05). The SSBBF-CTA technique exhibited excellent reproducibility and enabled accurate pretreatment diagnosis and assessment of spinal vascular malformations with high diagnostic efficiency, particularly for junior radiologists.

Deep learning for differentiation of osteolytic osteosarcoma and giant cell tumor around the knee joint on radiographs: a multicenter study

ObjectivesTo develop a deep learning (DL) model for differentiating between osteolytic osteosarcoma (OS) and giant cell tumor (GCT) on radiographs.MethodsPatients with osteolytic OS and GCT proven by postoperative pathology were retrospectively recruited from four centers (center A, training and internal testing; centers B, C, and D, external testing). Sixteen radiologists with different experiences in musculoskeletal imaging diagnosis were divided into three groups and participated with or without the DL model’s assistance. DL model was generated using EfficientNet-B6 architecture, and the clinical model was trained using clinical variables. The performance of various models was compared using McNemar’s test.ResultsThree hundred thirty-three patients were included (mean age, 27 years ± 12 [SD]; 186 men). Compared to the clinical model, the DL model achieved a higher area under the curve (AUC) in both the internal (0.97 vs. 0.77, p = 0.008) and external test set (0.97 vs. 0.64, p < 0.001). In the total test set (including the internal and external test sets), the DL model achieved higher accuracy than the junior expert committee (93.1% vs. 72.4%; p < 0.001) and was comparable to the intermediate and senior expert committee (93.1% vs. 88.8%, p = 0.25; 87.1%, p = 0.35). With DL model assistance, the accuracy of the junior expert committee was improved from 72.4% to 91.4% (p = 0.051).ConclusionThe DL model accurately distinguished osteolytic OS and GCT with better performance than the junior radiologists, whose own diagnostic performances were significantly improved with the aid of the model, indicating the potential for the differential diagnosis of the two bone tumors on radiographs.Critical relevance statementThe deep learning model can accurately distinguish osteolytic osteosarcoma and giant cell tumor on radiographs, which may help radiologists improve the diagnostic accuracy of two types of tumors.Key points• The DL model shows robust performance in distinguishing osteolytic osteosarcoma and giant cell tumor.• The diagnosis performance of the DL model is better than junior radiologists’.• The DL model shows potential for differentiating osteolytic osteosarcoma and giant cell tumor.Graphical

A multicenter clinical AI system study for detection and diagnosis of focal liver lesions

Early and accurate diagnosis of focal liver lesions is crucial for effective treatment and prognosis. We developed and validated a fully automated diagnostic system named Liver Artificial Intelligence Diagnosis System (LiAIDS) based on a diverse sample of 12,610 patients from 18 hospitals, both retrospectively and prospectively. In this study, LiAIDS achieved an F1-score of 0.940 for benign and 0.692 for malignant lesions, outperforming junior radiologists (benign: 0.830-0.890, malignant: 0.230-0.360) and being on par with senior radiologists (benign: 0.920-0.950, malignant: 0.550-0.650). Furthermore, with the assistance of LiAIDS, the diagnostic accuracy of all radiologists improved. For benign and malignant lesions, junior radiologists’ F1-scores improved to 0.936-0.946 and 0.667-0.680 respectively, while seniors improved to 0.950-0.961 and 0.679-0.753. Additionally, in a triage study of 13,192 consecutive patients, LiAIDS automatically classified 76.46% of patients as low risk with a high NPV of 99.0%. The evidence suggests that LiAIDS can serve as a routine diagnostic tool and enhance the diagnostic capabilities of radiologists for liver lesions.

Interpretable artificial intelligence-based app assists inexperienced radiologists in diagnosing biliary atresia from sonographic gallbladder images

BackgroundA previously trained deep learning-based smartphone app provides an artificial intelligence solution to help diagnose biliary atresia from sonographic gallbladder images, but it might be impractical to launch it in real clinical settings. This study aimed to redevelop a new model using original sonographic images and their derived smartphone photos and then test the new model’s performance in assisting radiologists with different experiences to detect biliary atresia in real-world mimic settings.MethodsA new model was first trained retrospectively using 3659 original sonographic gallbladder images and their derived 51,226 smartphone photos and tested on 11,410 external validation smartphone photos. Afterward, the new model was tested in 333 prospectively collected sonographic gallbladder videos from 207 infants by 14 inexperienced radiologists (9 juniors and 5 seniors) and 4 experienced pediatric radiologists in real-world mimic settings. Diagnostic performance was expressed as the area under the receiver operating characteristic curve (AUC).ResultsThe new model outperformed the previously published model in diagnosing BA on the external validation set (AUC 0.924 vs 0.908, P = 0.004) with higher consistency (kappa value 0.708 vs 0.609). When tested in real-world mimic settings using 333 sonographic gallbladder videos, the new model performed comparable to experienced pediatric radiologists (average AUC 0.860 vs 0.876) and outperformed junior radiologists (average AUC 0.838 vs 0.773) and senior radiologists (average AUC 0.829 vs 0.749). Furthermore, the new model could aid both junior and senior radiologists to improve their diagnostic performances, with the average AUC increasing from 0.773 to 0.835 for junior radiologists and from 0.749 to 0.805 for senior radiologists.ConclusionsThe interpretable app-based model showed robust and satisfactory performance in diagnosing biliary atresia, and it could aid radiologists with limited experiences to improve their diagnostic performances in real-world mimic settings.

Development and Validation of a Nomogram Based on Multimodality Ultrasonography Images for Differentiating Malignant from Benign American College of Radiology Thyroid Imaging, Reporting and Data System (TI-RADS) 3–5 Thyroid Nodules

ObjectivesTo develop and validate a predictive nomogram based on combined image features of gray-scale ultrasonography (US), elastosonography (ES), and contrast-enhanced US (CEUS) to differentiate malignant from benign ACR TI-RADS3-5 thyroid nodules. Materials and MethodsAmong 2767 thyroid nodules scanned by CEUS in * Hospital between April 2014 and November 2018, 669 nodules classified as ACR TI-RADS 3-5 were included with confirmed diagnosis and ES examination. 455 nodules were set as a training cohort and 214 as a validation cohort. Images were categorized as gray-scale US ACR TI-RADS3, TI-RADS4, TI-RADS5; ES patterns of ES-1, ES-2; and CEUS patterns of either heterogeneous hypoenhancement, concentric hypoenhancement, homogeneous hyper/isoenhancement, no-perfusion, hypoenhancement with sharp margin, island-like enhancement, or ring-like enhancement. Based on multivariate logistic regression analysis, a predictive nomogram model was developed and validated by receiver operating characteristic curve analysis. ResultsIn the training cohort, ACR TI-RADS4-5, ES-2, heterogeneous hypoenhancement, concentric hypoenhancement, and homogeneous hyper/isoenhancement were selected as predictors of malignancy by univariate logistic regression analysis. A predictive nomogram (combining indices of ACR TI-RADS, ES and CEUS) showed excellent predictive ability for differentiating malignant from benign lesions in the training cohort (AUC, 0.93 [95% CI 0.90-0.95]). The prediction nomogram model demonstrated a sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of 0.84, 0.88, 0.91 and 0.81.In the validation cohort, the AUC of the prediction nomogram model was significantly higher than those of the single modalities (P<0.005) . The AUCs of the validation cohort were 0.93 (95% CI 0.89-0.96) and 0.93 (95% CI 0.89-0.97) respectively for senior and junior radiologists. The prediction nomogram model has a sensitivity, specificity, PPV and NPV of 0.86, 0.87, 0.87 and 0.86. ConclusionsA predictive nomogram model combining ACR TI-RADS, ES, and CEUS showed potential clinical utility in differentiating malignant from benign ACR TI-RADS3-5 thyroid nodules.

MI-DenseCFNet: deep learning–based multimodal diagnosis models for Aureus and Aspergillus pneumonia

ObjectiveTo build and merge a diagnostic model called multi-input DenseNet fused with clinical features (MI-DenseCFNet) for discriminating between Staphylococcus aureus pneumonia (SAP) and Aspergillus pneumonia (ASP) and to evaluate the significant correlation of each clinical feature in determining these two types of pneumonia using a random forest dichotomous diagnosis model. This will enhance diagnostic accuracy and efficiency in distinguishing between SAP and ASP.MethodsIn this study, 60 patients with clinically confirmed SAP and ASP, who were admitted to four large tertiary hospitals in Kunming, China, were included. Thoracic high-resolution CT lung windows of all patients were extracted from the picture archiving and communication system, and the corresponding clinical data of each patient were collected.ResultsThe MI-DenseCFNet diagnosis model demonstrates an internal validation set with an area under the curve (AUC) of 0.92. Its external validation set demonstrates an AUC of 0.83. The model requires only 10.24s to generate a categorical diagnosis and produce results from 20 cases of data. Compared with high-, mid-, and low-ranking radiologists, the model achieves accuracies of 78% vs. 75% vs. 60% vs. 40%. Eleven significant clinical features were screened by the random forest dichotomous diagnosis model.ConclusionThe MI-DenseCFNet multimodal diagnosis model can effectively diagnose SAP and ASP, and its diagnostic performance significantly exceeds that of junior radiologists. The 11 important clinical features were screened in the constructed random forest dichotomous diagnostic model, providing a reference for clinicians.Clinical relevance statementMI-DenseCFNet could provide diagnostic assistance for primary hospitals that do not have advanced radiologists, enabling patients with suspected infections like Staphylococcus aureus pneumonia or Aspergillus pneumonia to receive a quicker diagnosis and cut down on the abuse of antibiotics.Key points• MI-DenseCFNet combines deep learning neural networks with crucial clinical features to discern between Staphylococcus aureus pneumonia and Aspergillus pneumonia.• The comprehensive group had an area under the curve of 0.92, surpassing the proficiency of junior radiologists.• This model can enhance a primary radiologist’s diagnostic capacity.

An integrated model incorporating deep learning, hand-crafted radiomics and clinical and US features to diagnose central lymph node metastasis in patients with papillary thyroid cancer

ObjectiveTo evaluate the value of an integrated model incorporating deep learning (DL), hand-crafted radiomics and clinical and US imaging features for diagnosing central lymph node metastasis (CLNM) in patients with papillary thyroid cancer (PTC).MethodsThis retrospective study reviewed 613 patients with clinicopathologically confirmed PTC from two institutions. The DL model and hand-crafted radiomics model were developed using primary lesion images and then integrated with clinical and US features selected by multivariate analysis to generate an integrated model. The performance was compared with junior and senior radiologists on the independent test set. SHapley Additive exPlanations (SHAP) plot and Gradient-weighted Class Activation Mapping (Grad-CAM) were used for the visualized explanation of the model.ResultsThe integrated model yielded the best performance with an AUC of 0.841. surpassing that of the hand-crafted radiomics model (0.706, p < 0.001) and the DL model (0.819, p = 0.26). Compared to junior and senior radiologists, the integrated model reduced the missed CLNM rate from 57.89% and 44.74–27.63%, and decreased the rate of unnecessary central lymph node dissection (CLND) from 29.87% and 27.27–18.18%, respectively. SHAP analysis revealed that the DL features played a primary role in the diagnosis of CLNM, while clinical and US features (such as extrathyroidal extension, tumour size, age, gender, and multifocality) provided additional support. Grad-CAM indicated that the model exhibited a stronger focus on thyroid capsule in patients with CLNM.ConclusionIntegrated model can effectively decrease the incidence of missed CLNM and unnecessary CLND. The application of the integrated model can help improve the acceptance of AI-assisted US diagnosis among radiologists.

Using multimodal ultrasound including full-time-series contrast-enhanced ultrasound cines for identifying the nature of thyroid nodules.

Based on the conventional ultrasound images of thyroid nodules, contrast-enhanced ultrasound (CEUS) videos were analyzed to investigate whether CEUS improves the classification accuracy of benign and malignant thyroid nodules using machine learning (ML) radiomics and compared with radiologists. The B-mode ultrasound (B-US), real-time elastography (RTE), color doppler flow images (CDFI) and CEUS cines of patients from two centers were retrospectively gathered. Then, the region of interest (ROI) was delineated to extract the radiomics features. Seven ML algorithms combined with four kinds of radiomics data (B-US, B-US + CDFI + RTE, CEUS, and B-US + CDFI + RTE + CEUS) were applied to establish 28 models. The diagnostic performance of ML models was compared with interpretations from expert and nonexpert readers. A total of 181 thyroid nodules from 181 patients of 64 men (mean age, 42 years +/- 12) and 117 women (mean age, 46 years +/- 12) were included. Adaptive boosting (AdaBoost) achieved the highest area under the receiver operating characteristic curve (AUC) of 0.89 in the test set among 28 models when combined with B-US + CDFI + RTE + CEUS data and an AUC of 0.72 and 0.66 when combined with B-US and B-US + CDFI + RTE data. The AUC achieved by senior and junior radiologists was 0.78 versus (vs.) 0.69 (p > 0.05), 0.79 vs. 0.64 (p < 0.05), and 0.88 vs. 0.69 (p < 0.05) combined with B-US, B-US+CDFI+RTE and B-US+CDFI+RTE+CEUS, respectively. With the addition of CEUS, the diagnostic performance was enhanced for all seven classifiers and senior radiologists based on conventional ultrasound images, while no enhancement was observed for junior radiologists. The diagnostic performance of ML models was similar to senior radiologists, but superior to those junior radiologists.

Differential diagnosis of benign and lung adenocarcinoma presenting as larger solid nodules and masses based on multiscale CT radiomics.

To develop a better radiomic model for the differential diagnosis of benign and lung adenocarcinoma lesions presenting as larger solid nodules and masses based on multiscale computed tomography (CT) radiomics. This retrospective study enrolled 205 patients with solid nodules and masses from Center 1 between January 2010 and February 2022 and Center 2 between January 2019 and February 2022. After applying the inclusion and exclusion criteria, we retrospectively enrolled 165 patients from two centers and assigned them to the training dataset (n = 115) or the test dataset (n = 50). Radiomics features were extracted from volumes of interest on CT images. A gradient boosting decision tree (GBDT) was used for data dimensionality reduction to perform the final feature selection. Four models were developed using clinical data, conventional imaging features and radiomics features, namely, the clinical and image model (CIM), the plain CT radiomics model (PRM), the enhanced CT radiomics model (ERM) and the combined model (CM). Model performance was evaluated to determine the best model for identifying benign and lung adenocarcinoma presenting as larger solid nodules and masses. In the training dataset, the areas under the curve (AUCs) for the CIM, PRM, ERM, and CM were 0.718, 0.806, 0.819, and 0.917, respectively. The differential diagnostic capability of the ERM was better than that of the PRM and the CIM. The CM was optimal. Intermediate and junior radiologists and respiratory physicians achieved improved obviously diagnostic results with the radiomics model. The senior radiologists showed slight improved diagnostic results after using the radiomics model. Radiomics may have the potential to be used as a noninvasive tool for the differential diagnosis of benign and lung adenocarcinoma lesions presenting as larger solid nodules and masses.