Introduction
Our client, a multinational pharmaceutical and biotechnology company headquartered in Hungary, operates the largest pharmaceutical research center in the region, reached out to HCLTech to use data science to build a solution that captures information at greater levels of specificity.
The Challenge
Lack of established technology
The client needed a way to quantify the properties of a mitochondrial network within neurons to enable more effective analysis of medications for various neurological diseases. They wanted a tool that could automatically identify healthy and damaged cells/tissue, as there was no existing tool for this purpose.
The Objective
Building a tool to assess the health of biological cells
The client identified three key deliverables for the project that would enable them to gain access to the information needed to optimize their products.
- The need to develop a methodology for the automatic and reliable detection of mitochondria in high-resolution 3D confocal microscopy images — a process commonly known as segmentation
- A way to quantify the properties of a mitochondrial network based on the segmentation
- Finally, the solution had to enable researchers to tell if the cells were healthy or damaged based on the properties of the mitochondrial network
Hence, there was a need to develop completely new toolsets to reach the desired goals.
The Solution
Quantify the properties of mitochondrial networks
The project took shape as an R&D proof of concept program involving a lot of experimentation. The HCLTech team used the client’s existing IT environment as a reference to develop a methodology that biologists (end users) can leverage without data science and programming knowledge. With this requirement in mind, the team designed a solution that would remain flexible enough to serve future needs by allowing biologists to make minor modifications.
- The team focused mainly on quantifying the properties of a mitochondrial network and using it to assess the health of the cells.
- We developed robust and interpretable algorithms using an ML model that would be capable of handling the “noisy” nature of the training data. The algorithm found patterns more typical in healthy or damaged mitochondrial networks.
- The team validated the results through a test dataset for a series of 3D images created especially for this project.
- Throughout the project, biologists from the client’s side contributed deep domain knowledge that was essential to the project’s success. They conducted all experiments and provided the 3D confocal microscopy images, pre-screening the samples so that algorithms could be trained on samples where the impairment was clear. They were also able to identify the targets that the new methodology would have to meet and evaluated the masks — binary versions of an image showing the location of only mitochondria in the image — and the different patterns that were learned from the data.
- The solution took shape as Python code that enables the calculation of the geometrical features of mitochondria and a decision tree model/machine learning algorithm that had learned the differences in the main patterns of healthy and damaged cells.
- The code and algorithm provided a suitably future-proof solution, as they did not require any manual setup before execution to classify new images. After the HCLTech team handed over the original code, the client’s team used a separate in-house data set to create the final model that the company would rely on for further research.
The Impact
Improved insights in product R&D
- In two months, with the client’s support, HCLTech created the first fully-functioning, validated 3D imagery machine learning toolset for quantifying the properties of a mitochondrial network and deciding if the cells are healthy or impaired.
- The team also saw promising results in creating an automatic system for detecting mitochondria in high-resolution 3D confocal microscopy images. While a larger dataset and more effective image post-processing would be necessary to fully realize this part of the solution and make it generally applicable, the methodology created represented a major advancement compared to earlier solutions when used on the client’s data.
- Compared to solutions based on 2D images, the new methodology did not require experts’ rules for classification but could instead learn patterns from data and utilize information previously unavailable to biologists.
- The client can now can apply the ML model on large image sets to automatically and effectively judge if a particular treatment had been successful and concluded that, this tool would be essential in evaluating the effects of drug candidates on neuronal mitochondrial networks.
