Application of Deep Convolutional Neural Networks for Discriminating Benign, Borderline, and Malignant Serous Ovarian Tumors From Ultrasound Images

Role of artificial intelligence applied to ultrasound in gynecology oncology: A systematic review

The aim of this paper was to explore the role of artificial intelligence (AI) applied to ultrasound imaging in gynecology oncology. Web of Science, PubMed, and Scopus databases were searched. All studies were imported to RAYYAN QCRI software. The overall quality of the included studies was assessed using QUADAS-AI tool. Fifty studies were included, of these 37/50 (74.0%) on ovarian masses or ovarian cancer, 5/50 (10.0%) on endometrial cancer, 5/50 (10.0%) on cervical cancer, and 3/50 (6.0%) on other malignancies. Most studies were at high risk of bias for subject selection (i.e., sample size, source, or scanner model were not specified; data were not derived from open-source datasets; imaging preprocessing was not performed) and index test (AI models was not externally validated) and at low risk of bias for reference standard (i.e., the reference standard correctly classified the target condition) and workflow (i.e., the time between index test and reference standard was reasonable). Most studies presented machine learning models (33/50, 66.0%) for the diagnosis and histopathological correlation of ovarian masses, while others focused on automatic segmentation, reproducibility of radiomics features, improvement of image quality, prediction of therapy resistance, progression-free survival, and genetic mutation. The current evidence supports the role of AI as a complementary clinical and research tool in diagnosis, patient stratification, and prediction of histopathological correlation in gynecological malignancies. For example, the high performance of AI models to discriminate between benign and malignant ovarian masses or to predict their specific histology can improve the diagnostic accuracy of imaging methods.

Developing a deep learning model for predicting ovarian cancer in Ovarian-Adnexal Reporting and Data System Ultrasound (O-RADS US) Category 4 lesions: A multicenter study

PURPOSE

To develop a deep learning (DL) model for differentiating between benign and malignant ovarian tumors of Ovarian-Adnexal Reporting and Data System Ultrasound (O-RADS US) Category 4 lesions, and validate its diagnostic performance.

METHODS

A retrospective analysis of 1619 US images obtained from three centers from December 2014 to March 2023. DeepLabV3 and YOLOv8 were jointly used to segment, classify, and detect ovarian tumors. Precision and recall and area under the receiver operating characteristic curve (AUC) were employed to assess the model performance.

RESULTS

A total of 519 patients (including 269 benign and 250 malignant masses) were enrolled in the study. The number of women included in the training, validation, and test cohorts was 426, 46, and 47, respectively. The detection models exhibited an average precision of 98.68% (95% CI: 0.95-0.99) for benign masses and 96.23% (95% CI: 0.92-0.98) for malignant masses. Moreover, in the training set, the AUC was 0.96 (95% CI: 0.94-0.97), whereas in the validation set, the AUC was 0.93(95% CI: 0.89-0.94) and 0.95 (95% CI: 0.91-0.96) in the test set. The sensitivity, specificity, accuracy, positive predictive value, and negative predictive values for the training set were 0.943,0.957,0.951,0.966, and 0.936, respectively, whereas those for the validation set were 0.905,0.935, 0.935,0.919, and 0.931, respectively. In addition, the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value for the test set were 0.925, 0.955, 0.941, 0.956, and 0.927, respectively.

CONCLUSION

The constructed DL model exhibited high diagnostic performance in distinguishing benign and malignant ovarian tumors in O-RADS US category 4 lesions.

Accuracy of machine learning in the preoperative identification of ovarian borderline tumors: a meta-analysis

AIM

The objective of this study is to explore the diagnostic value of machine learning (ML) in borderline ovarian tumors through meta-analysis.

METHODS

Pubmed, Embase, Web of Science, and Cochrane Library databases were comprehensively retrieved from database inception untill February 16, 2023. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was adopted to evaluate the risk of bias in the original studies. Sub-group analyses of ML were conducted according to clinical features and radiomics features. We separately discussed the discriminative value of ML for borderline vs benign and borderline vs malignant tumors.

RESULTS

Eighteen studies involving 12,778 subjects were included in our analysis. The modeling variables mainly consisted of radiomics features (n=13) and a small number of clinical features (n=5). When distinguishing between borderline and benign tumors, the ML model based on radiomic features achieved a c-index of 0.782 (95% CI: 0.732-0.831), sensitivity of 0.75 (95% CI: 0.67-0.82), and specificity of 0.75 (95% CI: 0.67-0.81) in the validation set. When distinguishing between borderline and malignant tumors, the ML model based on radiomic features achieved a c-index of 0.916 (95% CI: 0.891-0.940), sensitivity of 0.86 (95% CI: 0.78-0.91), and specificity of 0.88 (95% CI: 0.82-0.92) in the validation set. In addition, we analyzed the discriminatory ability of radiologists and found that their sensitivity was 0.26 (95% CI: 0.12-0.46) and specificity was 0.94 (95% CI: 0.90-0.97).

CONCLUSIONS

ML has tremendous potential in the preoperative diagnosis and differentiation of borderline ovarian tumors and may be more accurate than radiologists in diagnosing and differentiating borderline ovarian tumors.

Ultrasound image-based nomogram combining clinical, radiomics, and deep transfer learning features for automatic classification of ovarian masses according to O-RADS

Background

Accurate and rapid discrimination between benign and malignant ovarian masses is crucial for optimal patient management. This study aimed to establish an ultrasound image-based nomogram combining clinical, radiomics, and deep transfer learning features to automatically classify the ovarian masses into low risk and intermediate-high risk of malignancy lesions according to the Ovarian- Adnexal Reporting and Data System (O-RADS).

Methods

The ultrasound images of 1,080 patients with 1,080 ovarian masses were included. The training cohort consisting of 683 patients was collected at the South China Hospital of Shenzhen University, and the test cohort consisting of 397 patients was collected at the Shenzhen University General Hospital. The workflow included image segmentation, feature extraction, feature selection, and model construction.

Results

The pre-trained Resnet-101 model achieved the best performance. Among the different mono-modal features and fusion feature models, nomogram achieved the highest level of diagnostic performance (AUC: 0.930, accuracy: 84.9%, sensitivity: 93.5%, specificity: 81.7%, PPV: 65.4%, NPV: 97.1%, precision: 65.4%). The diagnostic indices of the nomogram were higher than those of junior radiologists, and the diagnostic indices of junior radiologists significantly improved with the assistance of the model. The calibration curves showed good agreement between the prediction of nomogram and actual classification of ovarian masses. The decision curve analysis showed that the nomogram was clinically useful.

Conclusion

This model exhibited a satisfactory diagnostic performance compared to junior radiologists. It has the potential to improve the level of expertise of junior radiologists and provide a fast and effective method for ovarian cancer screening.

Multi-modality deep learning model reaches high prediction accuracy in the diagnosis of ovarian cancer

We evaluated the diagnostic performance of a multimodal deep-learning (DL) model for ovarian mass differential diagnosis. This single-center retrospective study included 1,054 ultrasound (US)-detected ovarian tumors (699 benign and 355 malignant). Patients were randomly divided into training (n = 675), validation (n = 169), and testing (n = 210) sets. The model was developed using ResNet-50. Three DL-based models were proposed for benign-malignant classification of these lesions: single-modality model that only utilized US images; dual-modality model that used US images and menopausal status as inputs; and multi-modality model that integrated US images, menopausal status, and serum indicators. After 5-fold cross-validation, 210 lesions were tested. We evaluated the three models using the area under the curve (AUC), accuracy, sensitivity, and specificity. The multimodal model outperformed the single- and dual-modality models with 93.80% accuracy and 0.983 AUC. The Multimodal ResNet-50 DL model outperformed the single- and dual-modality models in identifying benign and malignant ovarian tumors.

PMFFNet: A hybrid network based on feature pyramid for ovarian tumor segmentation

Ovarian cancer is a highly lethal malignancy in the field of oncology. Generally speaking, the segmentation of ovarian medical images is a necessary prerequisite for the diagnosis and treatment planning. Therefore, accurately segmenting ovarian tumors is of utmost importance. In this work, we propose a hybrid network called PMFFNet to improve the segmentation accuracy of ovarian tumors. The PMFFNet utilizes an encoder-decoder architecture. Specifically, the encoder incorporates the ViTAEv2 model to extract inter-layer multi-scale features from the feature pyramid. To address the limitation of fixed window size that hinders sufficient interaction of information, we introduce Varied-Size Window Attention (VSA) to the ViTAEv2 model to capture rich contextual information. Additionally, recognizing the significance of multi-scale features, we introduce the Multi-scale Feature Fusion Block (MFB) module. The MFB module enhances the network's capacity to learn intricate features by capturing both local and multi-scale information, thereby enabling more precise segmentation of ovarian tumors. Finally, in conjunction with our designed decoder, our model achieves outstanding performance on the MMOTU dataset. The results are highly promising, with the model achieving scores of 97.24%, 91.15%, and 87.25% in mACC, mIoU, and mDice metrics, respectively. When compared to several Unet-based and advanced models, our approach demonstrates the best segmentation performance.

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

Background

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.

Methods

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.

Results

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.

Conclusion

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.

Machine learning and radiomics for segmentation and classification of adnexal masses on ultrasound

Ultrasound-based models exist to support the classification of adnexal masses but are subjective and rely upon ultrasound expertise. We aimed to develop an end-to-end machine learning (ML) model capable of automating the classification of adnexal masses. In this retrospective study, transvaginal ultrasound scan images with linked diagnoses (ultrasound subjective assessment or histology) were extracted and segmented from Imperial College Healthcare, UK (ICH development dataset; n = 577 masses; 1444 images) and Morgagni-Pierantoni Hospital, Italy (MPH external dataset; n = 184 masses; 476 images). A segmentation and classification model was developed using convolutional neural networks and traditional radiomics features. Dice surface coefficient (DICE) was used to measure segmentation performance and area under the ROC curve (AUC), F1-score and recall for classification performance. The ICH and MPH datasets had a median age of 45 (IQR 35-60) and 48 (IQR 38-57) years old and consisted of 23.1% and 31.5% malignant cases, respectively. The best segmentation model achieved a DICE score of 0.85 ± 0.01, 0.88 ± 0.01 and 0.85 ± 0.01 in the ICH training, ICH validation and MPH test sets. The best classification model achieved a recall of 1.00 and F1-score of 0.88 (AUC:0.93), 0.94 (AUC:0.89) and 0.83 (AUC:0.90) in the ICH training, ICH validation and MPH test sets, respectively. We have developed an end-to-end radiomics-based model capable of adnexal mass segmentation and classification, with a comparable predictive performance (AUC 0.90) to the published performance of expert subjective assessment (gold standard), and current risk models. Further prospective evaluation of the classification performance of this ML model against existing methods is required.

Deep learning in ovarian cancer diagnosis: a comprehensive review of various imaging modalities

Ovarian cancer poses a major worldwide health issue, marked by high death rates and a deficiency in reliable diagnostic methods. The precise and prompt detection of ovarian cancer holds great importance in advancing patient outcomes and determining suitable treatment plans. Medical imaging techniques are vital in diagnosing ovarian cancer, but achieving accurate diagnoses remains challenging. Deep learning (DL), particularly convolutional neural networks (CNNs), has emerged as a promising solution to improve the accuracy of ovarian cancer detection. This systematic review explores the role of DL in improving the diagnostic accuracy for ovarian cancer. The methodology involved the establishment of research questions, inclusion and exclusion criteria, and a comprehensive search strategy across relevant databases. The selected studies focused on DL techniques applied to ovarian cancer diagnosis using medical imaging modalities, as well as tumour differentiation and radiomics. Data extraction, analysis, and synthesis were performed to summarize the characteristics and findings of the selected studies. The review emphasizes the potential of DL in enhancing the diagnosis of ovarian cancer by accelerating the diagnostic process and offering more precise and efficient solutions. DL models have demonstrated their effectiveness in categorizing ovarian tissues and achieving comparable diagnostic performance to that of experienced radiologists. The integration of DL into ovarian cancer diagnosis holds the promise of improving patient outcomes, refining treatment approaches, and supporting well-informed decision-making. Nevertheless, additional research and validation are necessary to ensure the dependability and applicability of DL models in everyday clinical settings.

Artificial Intelligence in Ultrasound Diagnoses of Ovarian Cancer: A Systematic Review and Meta-Analysis

Ovarian cancer is the sixth most common malignancy, with a 35% survival rate across all stages at 10 years. Ultrasound is widely used for ovarian tumour diagnosis, and accurate pre-operative diagnosis is essential for appropriate patient management. Artificial intelligence is an emerging field within gynaecology and has been shown to aid in the ultrasound diagnosis of ovarian cancers. For this study, Embase and MEDLINE databases were searched, and all original clinical studies that used artificial intelligence in ultrasound examinations for the diagnosis of ovarian malignancies were screened. Studies using histopathological findings as the standard were included. The diagnostic performance of each study was analysed, and all the diagnostic performances were pooled and assessed. The initial search identified 3726 papers, of which 63 were suitable for abstract screening. Fourteen studies that used artificial intelligence in ultrasound diagnoses of ovarian malignancies and had histopathological findings as a standard were included in the final analysis, each of which had different sample sizes and used different methods; these studies examined a combined total of 15,358 ultrasound images. The overall sensitivity was 81% (95% CI, 0.80-0.82), and specificity was 92% (95% CI, 0.92-0.93), indicating that artificial intelligence demonstrates good performance in ultrasound diagnoses of ovarian cancer. Further prospective work is required to further validate AI for its use in clinical practice.

2D/3D ultrasound diagnosis of pediatric distal radius fractures by human readers vs artificial intelligence

Wrist trauma is common in children and generally requires radiography for exclusion of fractures, subjecting children to radiation and long wait times in the emergency department. Ultrasound (US) has potential to be a safer, faster diagnostic tool. This study aimed to determine how reliably US could detect distal radius fractures in children, to contrast the accuracy of 2DUS to 3DUS, and to assess the utility of artificial intelligence for image interpretation. 127 children were scanned with 2DUS and 3DUS on the affected wrist. US scans were then read by 7 blinded human readers and an AI model. With radiographs used as the gold standard, expert human readers obtained a mean sensitivity of 0.97 and 0.98 for 2DUS and 3DUS respectively. The AI model sensitivity was 0.91 and 1.00 for 2DUS and 3DUS respectively. Study data suggests that 2DUS is comparable to 3DUS and AI diagnosis is comparable to human experts.

Deep convolutional neural networks for multiple histologic types of ovarian tumors classification in ultrasound images

Objective

This study aimed to evaluate and validate the performance of deep convolutional neural networks when discriminating different histologic types of ovarian tumor in ultrasound (US) images.

Material and methods

Our retrospective study took 1142 US images from 328 patients from January 2019 to June 2021. Two tasks were proposed based on US images. Task 1 was to classify benign and high-grade serous carcinoma in original ovarian tumor US images, in which benign ovarian tumor was divided into six classes: mature cystic teratoma, endometriotic cyst, serous cystadenoma, granulosa-theca cell tumor, mucinous cystadenoma and simple cyst. The US images in task 2 were segmented. Deep convolutional neural networks (DCNN) were applied to classify different types of ovarian tumors in detail. We used transfer learning on six pre-trained DCNNs: VGG16, GoogleNet, ResNet34, ResNext50, DensNet121 and DensNet201. Several metrics were adopted to assess the model performance: accuracy, sensitivity, specificity, FI-score and the area under the receiver operating characteristic curve (AUC).

Results

The DCNN performed better in labeled US images than in original US images. The best predictive performance came from the ResNext50 model. The model had an overall accuracy of 0.952 for in directly classifying the seven histologic types of ovarian tumors. It achieved a sensitivity of 90% and a specificity of 99.2% for high-grade serous carcinoma, and a sensitivity of over 90% and a specificity of over 95% in most benign pathological categories.

Conclusion

DCNN is a promising technique for classifying different histologic types of ovarian tumors in US images, and provide valuable computer-aided information.

Analysis of computer-aided diagnostics in the preoperative diagnosis of ovarian cancer: a systematic review

OBJECTIVES

Different noninvasive imaging methods to predict the chance of malignancy of ovarian tumors are available. However, their predictive value is limited due to subjectivity of the reviewer. Therefore, more objective prediction models are needed. Computer-aided diagnostics (CAD) could be such a model, since it lacks bias that comes with currently used models. In this study, we evaluated the available data on CAD in predicting the chance of malignancy of ovarian tumors.

METHODS

We searched for all published studies investigating diagnostic accuracy of CAD based on ultrasound, CT and MRI in pre-surgical patients with an ovarian tumor compared to reference standards.

RESULTS

In thirty-one included studies, extracted features from three different imaging techniques were used in different mathematical models. All studies assessed CAD based on machine learning on ultrasound, CT scan and MRI scan images. Per imaging method, subsequently ultrasound, CT and MRI, sensitivities ranged from 40.3 to 100%; 84.6-100% and 66.7-100% and specificities ranged from 76.3-100%; 69-100% and 77.8-100%. Results could not be pooled, due to broad heterogeneity. Although the majority of studies report high performances, they are at considerable risk of overfitting due to the absence of an independent test set.

CONCLUSION

Based on this literature review, different CAD for ultrasound, CT scans and MRI scans seem promising to aid physicians in assessing ovarian tumors through their objective and potentially cost-effective character. However, performance should be evaluated per imaging technique. Prospective and larger datasets with external validation are desired to make their results generalizable.

AI diagnostic performance based on multiple imaging modalities for ovarian tumor: A systematic review and meta-analysis

Background

In recent years, AI has been applied to disease diagnosis in many medical and engineering researches. We aimed to explore the diagnostic performance of the models based on different imaging modalities for ovarian cancer.

Methods

PubMed, EMBASE, Web of Science, and Wanfang Database were searched. The search scope was all published Chinese and English literatures about AI diagnosis of benign and malignant ovarian tumors. The literature was screened and data extracted according to inclusion and exclusion criteria. Quadas-2 was used to evaluate the quality of the included literature, STATA 17.0. was used for statistical analysis, and forest plots and funnel plots were drawn to visualize the study results.

Results

A total of 11 studies were included, 3 of them were modeled based on ultrasound, 6 based on MRI, and 2 based on CT. The pooled AUROCs of studies based on ultrasound, MRI and CT were 0.94 (95% CI 0.88-1.00), 0.82 (95% CI 0.71-0.93) and 0.82 (95% Cl 0.78-0.86), respectively. The values of I² were 99.92%, 99.91% and 92.64% based on ultrasound, MRI and CT. Funnel plot suggested no publication bias.

Conclusion

The models based on ultrasound have the best performance in diagnostic of ovarian cancer.

Automatic ovarian tumors recognition system based on ensemble convolutional neural network with ultrasound imaging

Background

Upon the discovery of ovarian cysts, obstetricians, gynecologists, and ultrasound examiners must address the common clinical challenge of distinguishing between benign and malignant ovarian tumors. Numerous types of ovarian tumors exist, many of which exhibit similar characteristics that increase the ambiguity in clinical diagnosis. Using deep learning technology, we aimed to develop a method that rapidly and accurately assists the different diagnosis of ovarian tumors in ultrasound images.

Methods

Based on deep learning method, we used ten well-known convolutional neural network models (e.g., Alexnet, GoogleNet, and ResNet) for training of transfer learning. To ensure method stability and robustness, we repeated the random sampling of the training and validation data ten times. The mean of the ten test results was set as the final assessment data. After the training process was completed, the three models with the highest ratio of calculation accuracy to time required for classification were used for ensemble learning pertaining. Finally, the interpretation results of the ensemble classifier were used as the final results. We also applied ensemble gradient-weighted class activation mapping (Grad-CAM) technology to visualize the decision-making results of the models.

Results

The highest mean accuracy, mean sensitivity, and mean specificity of ten single CNN models were 90.51 ± 4.36%, 89.77 ± 4.16%, and 92.00 ± 5.95%, respectively. The mean accuracy, mean sensitivity, and mean specificity of the ensemble classifier method were 92.15 ± 2.84%, 91.37 ± 3.60%, and 92.92 ± 4.00%, respectively. The performance of the ensemble classifier is better than that of a single classifier in three evaluation metrics. Moreover, the standard deviation is also better which means the ensemble classifier is more stable and robust.

Conclusion

From the comprehensive perspective of data quantity, data diversity, robustness of validation strategy, and overall accuracy, the proposed method outperformed the methods used in previous studies. In future studies, we will continue to increase the number of authenticated images and apply our proposed method in clinical settings to increase its robustness and reliability.

Artificial intelligence performance in image-based ovarian cancer identification: A systematic review and meta-analysis

BACKGROUND

Accurate identification of ovarian cancer (OC) is of paramount importance in clinical treatment success. Artificial intelligence (AI) is a potentially reliable assistant for the medical imaging recognition. We systematically review articles on the diagnostic performance of AI in OC from medical imaging for the first time.

METHODS

The Medline, Embase, IEEE, PubMed, Web of Science, and the Cochrane library databases were searched for related studies published until August 1, 2022. Inclusion criteria were studies that developed or used AI algorithms in the diagnosis of OC from medical images. The binary diagnostic accuracy data were extracted to derive the outcomes of interest: sensitivity (SE), specificity (SP), and Area Under the Curve (AUC). The study was registered with the PROSPERO, CRD42022324611.

FINDINGS

Thirty-four eligible studies were identified, of which twenty-eight studies were included in the meta-analysis with a pooled SE of 88% (95%CI: 85-90%), SP of 85% (82-88%), and AUC of 0.93 (0.91-0.95). Analysis for different algorithms revealed a pooled SE of 89% (85-92%) and SP of 88% (82-92%) for machine learning; and a pooled SE of 88% (84-91%) and SP of 84% (80-87%) for deep learning. Acceptable diagnostic performance was demonstrated in subgroup analyses stratified by imaging modalities (Ultrasound, Magnetic Resonance Imaging, or Computed Tomography), sample size (≤300 or >300), AI algorithms versus clinicians, year of publication (before or after 2020), geographical distribution (Asia or non Asia), and the different risk of bias levels (≥3 domain low risk or < 3 domain low risk).

INTERPRETATION

AI algorithms exhibited favorable performance for the diagnosis of OC through medical imaging. More rigorous reporting standards that address specific challenges of AI research could improve future studies.

FUNDING

This work was supported by the Natural Science Foundation of China (No. 82073647 to Q-JW and No. 82103914 to T-TG), LiaoNing Revitalization Talents Program (No. XLYC1907102 to Q-JW), and 345 Talent Project of Shengjing Hospital of China Medical University (No. M0268 to Q-JW and No. M0952 to T-TG).