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

Abstract

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.

Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology

“…The study by Gao et al used a large, multicenter, and heterogenous data set, which disclosed that AI-enabled US outperformed an average trained radiologist in discriminat-ing malignant and benign ovarian masses and improved the examiner's accuracy [189]. These findings were consistent with other studies [185,186], but there were also studies with smaller sample sizes that showed a level of performance reaching those of human experts [188,191,192].…”

“…Of the 11 extracted articles, only two were designed as prospective studies [184,185]. Included studies were published from 2009-2023, whereby the research group Amor et al was the first to describe AI application in sonographic assessment using a non-specified pattern recognition analysis to classify adnexal masses in a new reporting system [184].…”

“…Enabling an automated discrimination between benign and malignant tumors was a predominant focus of the current research, represented in six studies included [185][186][187][188][189][190]. Three studies assessed the performance of automated tumor classification [184,191,192], one study developed a population-based screening method for BRCA mutations [193], and one study focused on the automated elimination of artefacts and objects in US images to increase the accuracy of the AI model [194].…”

“…Three studies assessed the performance of automated tumor classification [184,191,192], one study developed a population-based screening method for BRCA mutations [193], and one study focused on the automated elimination of artefacts and objects in US images to increase the accuracy of the AI model [194]. Aramendía-Vidaurreta et al was the only group to investigate the automated discrimination of benign and malignant masses in 3D US images [187] and Hsu et al distinguished between transabdominal and transvaginal US images [185].…”

Equipping Computational Pathology Systems with Artifact Processing Pipelines: A Showcase for Computation and Performance Trade-offs

“…To address these problems, ensembles of DL models (a.k.a. deep ensembles) have been used to overcome the weakness of an individual model [28][29][30]. Ensemble methods combine the prediction of independent models using averaging or majority voting.…”