Application of Deep Learning Model in the Sonographic Diagnosis of Uterine Adenomyosis

Abstract

Background: This study aims to evaluate the diagnostic performance of Deep Learning (DL) machine for the detection of adenomyosis on uterine ultrasonographic images and compare it to intermediate ultrasound skilled trainees. Methods: Prospective observational study were conducted between 1 and 30 April 2022. Transvaginal ultrasound (TVUS) diagnosis of adenomyosis was investigated by an experienced sonographer on 100 fertile-age patients. Videoclips of the uterine corpus were recorded and sequential ultrasound images were extracted. Intermediate ultrasound-skilled trainees and DL machine were asked to make a diagnosis reviewing uterine images. We evaluated and compared the accuracy, sensitivity, positive predictive value, F1-score, specificity and negative predictive value of the DL model and the trainees for adenomyosis diagnosis. Results: Accuracy of DL and intermediate ultrasound-skilled trainees for the diagnosis of adenomyosis were 0.51 (95% CI, 0.48–0.54) and 0.70 (95% CI, 0.60–0.79), respectively. Sensitivity, specificity and F1-score of DL were 0.43 (95% CI, 0.38–0.48), 0.82 (95% CI, 0.79–0.85) and 0.46 (0.42–0.50), respectively, whereas intermediate ultrasound-skilled trainees had sensitivity of 0.72 (95% CI, 0.52–0.86), specificity of 0.69 (95% CI, 0.58–0.79) and F1-score of 0.55 (95% CI, 0.43–0.66). Conclusions: In this preliminary study DL model showed a lower accuracy but a higher specificity in diagnosing adenomyosis on ultrasonographic images compared to intermediate-skilled trainees.

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

“…Further fields of applications were found in the process of this literature review. In total, 11 articles were summarized in this section, including the topics of endometriosis [216,217], premature ovarian failure [218,219], uterine fibroids [220,221], follicle tracking [222,223], and ectopic pregnancies [224,225]. Another study addressed the issue of poor image quality in 3D US images due to data processing and showed that AI image enhancement methods could produce increased 3D image quality with user-preferential flexibility in both gynecological and obstetric US images [226].…”

“…Two included articles discussed the sensitive topic of endometriosis, which can be problematic for both physician and patient due to complex clinical management and impaired quality of life in affected women [216,217]. Both studies had the usage of transvaginal 2D US videos and the missing histopathological or surgical confirmation in common, but focused on two different manifestations of endometriosis.…”

“…Maicas et al developed a highly accurate AI model for the classification of the pouch of Douglas obliteration as a cause of pelvic inflammation often seen in endometriosis via detection of the so-called 'sliding sign' [216]. In comparison, the results of Raimondo et al showed a low sensitivity of the AI model to detect adenomyosis, but a high specificity, interpreted as a useful tool to rather exclude than detect adenomyosis [217].…”

Automated Uterine Fibroids Detection in Ultrasound Images Using Deep Convolutional Neural Networks

“…One of the most frequent gynecological disorders affecting women of childbearing age is UF [26]. Some of the signs of these benign tumors of the uterine muscle are painful periods, inability to conceive, and heavy monthly bleeding [27].…”

“…Similarly, a study [17] achieved 96.4% accuracy using the VGG16 CNN model as a pre-trained feature extractor, which was trained on their ultrasound images dataset. To automate UF detection from an ultrasound image, a DCNN architecture was developed in another work [26]. The suggested model extracted features from ultrasound images using a combination of convolutional, pooling, and fully connected layers.…”