Automatic segmentation method of pelvic floor levator hiatus in ultrasound using a self-normalizing neural network

Abstract

Tom Vercauteren, "Automatic segmentation method of pelvic floor levator hiatus in ultrasound using a selfnormalizing neural network," J. Med. Imag. 5(2), 021206 (2018), doi: 10.1117/1.JMI.5.2.021206. Abstract. Segmentation of the levator hiatus in ultrasound allows the extraction of biometrics, which are of importance for pelvic floor disorder assessment. We present a fully automatic method using a convolutional neural network (CNN) to outline the levator hiatus in a two-dimensional image extracted from a three-dimensional ultrasound volume. In particular, our method uses a recently developed scaled exponential linear unit (SELU) as a nonlinear self-normalizing activation function, which for the first time has been applied in medical imaging with CNN. SELU has important advantages such as being parameter-free and mini-batch independent, which may help to overcome memory constraints during training. A dataset with 91 images from 35 patients during Valsalva, contraction, and rest, all labeled by three operators, is used for training and evaluation in a leave-one-patient-out cross validation. Results show a median Dice similarity coefficient of 0.90 with an interquartile range of 0.08, with equivalent performance to the three operators (with a Williams' index of 1.03), and outperforming a U-Net architecture without the need for batch normalization. We conclude that the proposed fully automatic method achieved equivalent accuracy in segmenting the pelvic floor levator hiatus compared to a previous semiautomatic approach.

Deep learning enables automatic quantitative assessment of puborectalis muscle and urogenital hiatus in plane of minimal hiatal dimensions

“…Our results can be compared to the (semi) automatic segmentation results obtained for the UH, reported previously by Sindhwani et al . and Bonmati et al …”

“…In medicine, CNNs are used for diagnosis, for example in discrimination between images of benign and malignant skin lesions and the detection of Alzheimer's disease on magnetic resonance imaging (MRI). Segmentation can also be learned by CNN and was found to be successful in, for example, segmenting brain structures on MRI and ultrasound and segmentation of the urogenital hiatus (UH).…”

“…These parameters are measured manually, making them observer dependent and time consuming, limiting their introduction into general practice. To automate these measurements, Bonmati et al . used a CNN for UH segmentation.…”

Convolutional neural network‐based pelvic floor structure segmentation using magnetic resonance imaging in pelvic organ prolapse

“…There are some differences with previous investigations. [35][36][37][38][39][40] Different modalities of medical imaging techniques have their own advantages. Two groups used ultrasound images to accomplish levator hiatus segmentation using the fully CNN (FCN) and U-Net.…”

“…Two groups used ultrasound images to accomplish levator hiatus segmentation using the fully CNN (FCN) and U-Net. 37,41 Wang et al 38 and He et al 39 investigated prostate, rectum and bladder segmentation using axial view computed tomography based on a multistage FCN. Techniques including dilated convolution 42 and full-resolution residual network 43 were also investigated to deal with the blurry edges of objects by capturing a larger field of view information.…”