Trophectoderm segmentation in human embryo images via inceptioned U-Net.

Trophectoderm (TE) is one of the main components of a day-5 human embryo (blastocyst) that correlates with the embryo's quality. Precise segmentation of TE is an important step toward achieving automatic human embryo quality assessment based on morphological image features. Automatic segmentation of TE, however, is a challenging task and previous work on this is quite limited. In this paper, four fully convolutional deep models are proposed for accurate segmentation of trophectoderm in microscopic images of the human blastocyst. In addition, a multi-scaled ensembling method is proposed that aggregates five models trained at various scales offering trade-offs between the quantity and quality of the spatial information. Furthermore, synthetic embryo images are generated for the first time to address the lack of data in training deep learning models. These synthetically generated images are proven to be effective to fill the generalization gap in deep learning when limited data is available for training. Experimental results confirm that the proposed models are capable of segmenting TE regions with an average Precision, Recall, Accuracy, Dice Coefficient and Jaccard Index of 83.8%, 90.1%, 96.9%, 86.61% and 76.71%, respectively. Particularly, the proposed Inceptioned U-Net model outperforms state-of-the-art by 10.3% in Accuracy, 9.3% in Dice Coefficient and 13.7% in Jaccard Index. Further experiments are conducted to highlight the effectiveness of the proposed models compared to some recent deep learning based segmentation methods.

Predicting Human Embryos' Implantation Outcome from a Single Blastocyst Image.

Only one-third of embryo transfer cycles via invitro fertilization, the most common fertility treatment, leads to a clinical pregnancy. Identifying embryos with the highest potentials for transfer is an essential step to optimize in-vitro fertilization outcome. However, human embryos are complicated by nature and some of their developmental aspects has still remained a mystery to expert biologists. In this paper, the first-ever attempt is made to estimate probability of implantation using a single blastocyst image. First, a semantic segmentation system is proposed for human blastocyst components in microscopic images. Second, a multi-stream classification model is proposed for the prediction of embryos' implantation outcome. The proposed classification model features an architectural component, Compact-Contextualize-Calibrate (C3) to guide the feature extraction process and a slow-fusion strategy to learn cross-modality features. Experimental results confirm that the proposed method delivers the first-reported implantation outcome prediction via a single blastocyst image to date with a mean accuracy of 70.9%.

A hybrid approach for multiple blastomeres identification in early human embryo images.

Automatic quality assessment of the human embryo paves the way to improve the outcome of the In Vitro Fertilization (IVF) treatment by selecting embryos with the highest implantation potentials. Analyzing the shape, size, and motion of the cells, as well as other time-related changes, facilitates embryo quality assessment. However, the ambitious 3D-like side-lit appearance of the embryo, occlusion, transparency of cells and artifacts such as fragmentation make automatic detection of blastomeres (embryonic cells) a challenging task. In this paper, an automated noninvasive approach is proposed to identify multiple blastomere cells inside an embryo at different growth stages. In particular, the proposed method aims to identify up to 8 blastomeres in microscopic human embryo images of days 1-3. The proposed system is a hybrid approach that aggregates both models and features capturing global and local characteristics to locate the boundaries of each blastomere. Experimental results on a large dataset of 271 embryo images with various blastomere numbers and sizes confirm that the proposed method identifies blastomeres with average Precision, Recall, and Overall Quality of 85.9%, 85.3%, and 76.5%, respectively.

A unified data embedding and scrambling method.

Conventionally, data embedding techniques aim at maintaining high-output image quality so that the difference between the original and the embedded images is imperceptible to the naked eye. Recently, as a new trend, some researchers exploited reversible data embedding techniques to deliberately degrade image quality to a desirable level of distortion. In this paper, a unified data embedding-scrambling technique called UES is proposed to achieve two objectives simultaneously, namely, high payload and adaptive scalable quality degradation. First, a pixel intensity value prediction method called checkerboard-based prediction is proposed to accurately predict 75% of the pixels in the image based on the information obtained from 25% of the image. Then, the locations of the predicted pixels are vacated to embed information while degrading the image quality. Given a desirable quality (quantified in SSIM) for the output image, UES guides the embedding-scrambling algorithm to handle the exact number of pixels, i.e., the perceptual quality of the embedded-scrambled image can be controlled. In addition, the prediction errors are stored at a predetermined precision using the structure side information to perfectly reconstruct or approximate the original image. In particular, given a desirable SSIM value, the precision of the stored prediction errors can be adjusted to control the perceptual quality of the reconstructed image. Experimental results confirmed that UES is able to perfectly reconstruct or approximate the original image with SSIM value > 0.99 after completely degrading its perceptual quality while embedding at 7.001 bpp on average.