Deep Bv: A Fully Automated System for Brain Ventricle Localization and Segmentation In 3D Ultrasound Images of Embryonic Mice

Abstract

Volumetric analysis of brain ventricle (BV) structure is a key tool in the study of central nervous system development in embryonic mice. High-frequency ultrasound (HFU) is the only non-invasive, real-time modality available for rapid volumetric imaging of embryos in utero. However, manual segmentation of the BV from HFU volumes is tedious, time-consuming, and requires specialized expertise. In this paper, we propose a novel deep learning based BV segmentation system for whole-body HFU images of mouse embryos. Our fully automated system consists of two modules: localization and segmentation. It first applies a volumetric convolutional neural network on a 3D sliding window over the entire volume to identify a 3D bounding box containing the entire BV. It then employs a fully convolutional network to segment the detected bounding box into BV and background. The system achieves a Dice Similarity Coefficient (DSC) of 0.8956 for BV segmentation on an unseen 111 HFU volume test set surpassing the previous state-of-the-art method (DSC of 0.7119) by a margin of 25%.

Automatic Mouse Embryo Brain Ventricle & Body Segmentation and Mutant Classification From Ultrasound Data Using Deep Learning

“…The system was trained on 259 HFU images with manual BV segmentation and achieved a DSC score of 0.896 on the unseen 111 volume test set. More details and sample results can be found in [8].…”

“…The research described in this paper was supported in part by NIH grant EB022950. Inspired by the success of the work in [8], our framework was extended to segment the body surface using a deep-learning based framework. The challenges for body segmentation are similar to those for BV segmentation, except that the imbalance between the background and the foreground is not as extreme (i.e., the body makes up around 10% of the whole volume on average).…”

Deep Learning for Carotid Plaque Segmentation using a Dilated U-Net Architecture

“…This was possible because similar errors were also incorporated in training which ensured that the network could handle these errors. Qiu et al 13 and Amiri et al 30 utilized another CNN to identify the correct bounding box prior to performing the segmentation. This would be challenging in our diverse dataset as we observe from Figure 7(a) where the automatic U-Net could not even detect the correct location of the plaque in the B-mode image.…”

“…[12][13][14][15][16][17][18] Use of three-dimensional (3D) data sets provide significant improvements when compared to 2D data sets. 12,13,18 Qiu et al 13 developed a fully automatic segmentation method using CNN for 3D segmentation of the brain ventricle structure in embryonic mice. Their implementation consisted of a two-stage process where the first stage performed localization of the relevant structure within a bounding box, which was then supplied as an input to the second stage, which performed the segmentation.…”

3D Bounding Box Detection in Volumetric Medical Image Data: A Systematic Literature Review

“…Relying on the fact that there is at most one instance of a organ, BBs with the same label are merged into one. [14] scan the whole volume using a 3D sliding window, that is large enough to fully contain the target structure. A 10-layer VGGNet [13] serves as the classifier.…”