AI Powered Quantitive Pathology in Clinical and Preclinical StudiesWatch now
- Solutions by application
- Explore some of our use cases!
Huntington's disease (HD) is a progressive brain disorder caused by a single defective gene, called huntingtin. Over time this leads to brain damage that causes symptoms such as abnormal involuntary movements, severe cognitive decline, and mood changes. As a dominant genetic defect, anyone who inherits it will eventually develop the disease and there is currently no cure. Instead, treatments focus on managing symptoms but the progression of HD eventually leads to complete dependency on full-time care.
Current models for predicting HD development are imprecise and unreliable at an individual level. While there are genetic tests, these detection methods account for only half of HD patients. This has pushed neuroimaging for biomarkers, such as mutant huntingtin (mHtt), to become more common. However, traditional methods for image analysis are time-consuming and have high variability between pathologists and laboratories.
Improving mHtt detection methods is crucial to reliably predicting clinical diagnosis of HD and developing new therapies. Artificial Intelligence (AI)-assisted image analysis increases speed and scalability of projects, saving researchers valuable time to focus their efforts on more complex and higher level challenges. Automated methods lead to improved statistical accuracy and reproducible results for future HD research.
Doctoral student Polina Stepanova is implementing AI models in her Huntington’s disease research, using Aiforia Create to analyze mutant huntingtin (mHtt) aggregation in brain tissue. We interviewed her about this project and her experience working with Aiforia’s software.
Polina: The Aiforia software was important for our research work. The project is linked with a novel therapeutic approach for Huntington's disease. Aiforia’s software was used to analyze the volume of mutant huntingtin (mHtt) aggregation in brain tissue.
Nowadays, there is no perfect detection system for mutant huntingtin (mHtt) aggregation, which is involved in the pathophysiology of Huntington's disease. However, it is crucial to have a trustful detection mechanism for this parameter. The vast majority of scientific articles regarding this research question do not show the final analysis of the mHtt aggregation and only demonstrate representative pictures, which does not give an accurate outcome.
Firstly, Aiforia allows to decrease the bias of the researcher in the analysis. Therefore, the outcome is more trustful data, which is essential in scientific work.
Secondly, Aiforia software gives more accurate cell detection in comparison with other available software for counting cells and aggregates.
Third, more data can be processed during a short time period with Aiforia’s software.
I used less than 15 images to train neural networks of my AI model, about 900 annotations for cell detection and about 300 annotations for nuclear inclusions detection. However, I believe that it could be even less; everything depends on the quality and similarity of the samples for detection.
The most significant benefit of Aiforia to my work in neuroscience and the science in general is unbiased, truthful analysis. Moreover, a wide range of tools and parameters can be obtained as an outcome, making Aiforia software an essential and valuable tool.