Digital Image Analysis with Fully Connected Convolutional Neural Network to Facilitate Hysteroscopic Fibroid Resection

Abstract

Aims: The study aimed to determine the accuracy of deep neural network in identifying the plane between myoma and normal myometrium. Methods: On the images of surgery, different structures were signed and annotated for the training phase. After the appropriate training of the deep neural network with 4,688 images from that training set, 1,600 formerly unseen images were used for testing. Indication for surgery was heavy menstrual bleeding and hysteroscopic finding was submucous fibroid. Operative intervention was fibroid resection. Recorded videos of transcervical resection of myoma in 13 cases were used for the study. Different filters and procedures were applied by the fully convolutional neural network (FCNN) for identifying previously annotated structures. Results: Previously manually annotated images and the manually drawn bitmasks were used for training the applied FCNN and then this pre-trained network was used for automatic segmentation of normal myometrium in an unseen video frame. The segmentation pixel-wise accuracy achieved the 86.19% considering the Hausdorff metric. Conclusion: Using deep learning technique in analyzing process of endoscopic video frame could help in real-time identification of structures while performing endoscopic surgery.

Automated Detection of Endometrial Polyps from Hysteroscopic Videos Using Deep Learning

“…The current use of deep learning in hysteroscopic images is mainly focused on the classification of endometrial cancer, with a few studies on endometrial fibroids [ 16 , 17 ]. For example, Török et al used a fully convolutional CNN to identify the plane between myoma and the normal myometrium, achieving a pixel-wise segmentation accuracy of 86.19% after training the network on 13 cases of video data for 140 epochs [ 22 ]. Deep learning research on uterine lesions has mainly focused on MRIs and ultrasound images.…”

New Hysteroscopic Approaches to Uterine Fibroids

“…An important help for better assessing the pseudocapsule could be brought by computer-aided imaging. A recent study of Török et al [6] showed a very good accuracy, of more than 86%, using fully convolutional neural networking and highresolution endoscopic image, which is promising for faster and safer future resections.…”

Bridging the Diagnostic Gap between Histopathologic and Hysteroscopic Chronic Endometritis with Deep Learning Models

“…Some research groups have introduced CNN models into the image analysis of hysteroscopy for the diagnosis of endometrial pathologic conditions. Török et al [ 81 ] first reported the application of the fully connected CNN model in the identification of the boundaries between the intact myometrium and uterine submucosal fibroid lesion in hysteroscopic surgery. They trained their model with 4688 hysteroscopic images and tested it with 1600 formerly unseen images, achieving an accuracy of 86.19% in hysteroscopic surgery for women with heavy menstrual bleeding and submucosal uterine fibroids.…”