Update paper.md

Author: devoworm

paper.md

@@ -32,7 +32,7 @@ affiliations:
 ---  
 
 # Summary
-Extracting metadata from microscopic videos/images have been one of the key steps in the process of finding emerging patterns from various biological processes. There have been many attempts to develop segmentation tools for cell shape and location (Cao, 2019A; Cao, 2019B; Chen, 2013). In particular, cell tracking methodologies provide quantitative summaries of cell centroid positions within an embryo (Ulman, 2006). Our pre-trained models (Devolearn) aim to speed up this process of collecting metadata by using robust deep learning models that can be used through a high level API. Devolearn’s primary focus is the _Caenorhabditis elegans_ embryo and specifically on the early embryogenesis process. This builds upon desired functionality that was first proposed by the DevoWorm group in (Alicea, 2019). Below are some of the capabilities of the DevoLearn model.
+Extracting metadata from microscopic videos/images have been one of the key steps in the process of finding emerging patterns from various biological processes. There have been many attempts to develop segmentation tools for cell shape and location (Cao, 2019a; Cao, 2019b; Chen, 2013). In particular, cell tracking methodologies provide quantitative summaries of cell centroid positions within an embryo (Ulman, 2006). Our pre-trained models (Devolearn) aim to speed up this process of collecting metadata by using robust deep learning models that can be used through a high level API. Devolearn’s primary focus is the _Caenorhabditis elegans_ embryo and specifically on the early embryogenesis process. This builds upon desired functionality that was first proposed by the DevoWorm group in (Alicea, 2019). Below are some of the capabilities of the DevoLearn model.
 
 * **Segments images/videos of the _C. elegans_ embryo** and extract the centroids of the cells and save them into a CSV file.  
 
@@ -43,7 +43,7 @@ Extracting metadata from microscopic videos/images have been one of the key step
 DevoLearn has been made available as an open-source module, available on PyPI ([https://pypi.org/project/devolearn/](https://pypi.org/project/devolearn/)). All the deep-learning models used in devolearn are built and trained on PyTorch. The PyPI package itself does not contain the model weights, but the models are downloaded automatically once the user imports a certain model from the package. 
 
 ## Technical Details  
-DevoLearn 0.2.0 is optimized to segment and analyze high-resolution microscopy images such as those acquired using light sheet microscopy. The deep learning models used for embryo segmentation and cell lineage population prediction were both based on the ResNet18 architecture. Data from the EPIC dataset (Murray, 2012) was used to train the GAN (beta) and the lineage wise cell population prediction model. The embryo segmentation model was trained on a dataset sourced from [this paper](https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-019-2720-x#Bib1). 
+DevoLearn 0.2.0 is optimized to segment and analyze high-resolution microscopy images such as those acquired using light sheet microscopy. The deep learning models used for embryo segmentation and cell lineage population prediction were both based on the ResNet18 architecture. Data from the EPIC dataset (Murray, 2012) was used to train the GAN (beta) and the lineage wise cell population prediction model. The embryo segmentation model was trained on a dataset sourced from Cao (2019b).
 
 ## Statement of Need
 Devolearn (0.2.0) is a Python package that aims to automate the process of collecting metadata from videos/images of the _C. elegans_ embryo with the help of deep learning models \autoref{fig:1}. This would enable researchers/enthusiasts to analyse features from videos/images at scale without having to annotate their data manually. There are a number of pre-trained models which are already in use in different contexts, but options are fewer within the unique feature space of developmental biology, in particular. Devolearn aims not just to fix this issue, but also work on other aspects around developmental biology with species-specific models.