This is an example spatial distributions of ADC values in selected tumors. Visually, HCT116 tumors are more complex and variable than LoVo tumors. Credit: Paul Tar
Scientists are using machine learning to take some of the uncertainty out of how cancer treatments are actually impacting tumors.
Researchers from the University of Manchester are using a machine learning method called Linear Poisson Modelling—developed to help scientists identify craters and dunes of Mars—as a new way to measure the effects of cancer treatments.
Linear Poisson Modelling works by learning patterns within data and how they can change, enabling researchers to assess the effects of errors in data, and providing additional output predictions of how precise its results are.
“This technique is all about making the most of ‘small data’, which is common in medical studies where it is difficult to obtain large numbers of samples,” Paul Tar, PhD, who co-developed the method during his PhD project, said in a statement. “Researchers use charitable or public money, so it is important that they use it in the most efficient way possible, something which this technique allows.”
Researchers often have difficulty understanding what effect treatment is having on a tumor against a background of changes that would have occurred regardless of treatment, as tumors are not uniform and different parts of the tumor change at different speeds.
To obtain meaningful results, researchers often look at average changes in tumors using several samples, often with animals. However, this makes it difficult to assess the effects of treatment on each individual.
“Every person’s cancer is unique, which can make treating the disease challenging as a drug that works for one patient might not work for someone else,” James O’Connor, PhD, a Cancer Research UK advanced clinician scientist, said in a statement. “That’s why we’re increasingly looking at finding new ways to make treatment more personal, and this innovative work could be a step towards that goal. The next step will be further research to find out if that’s the case, and to help uncover this method’s potential.”
The new technique was first developed to help planetary scientists map features of distant planets like Mars and better understand the errors and uncertainties of observations.
The Manchester researchers applied the technique to tumor samples and were able to demonstrate a four-fold increase in the precision of tumor change measurements that detected the beneficial effects of cancer therapies.
“The results of this study show that we can present findings which researchers can be much more certain of,” Neil Thacker, PhD, from the University’s Division of Informatics, Imaging & Data Sciences, said in a statement. “This means you can get the same quality of data from one sample instead of 16.
“This has important implications for research, meaning that instead of using 16 mice, in some studies only one is needed,” he added. “This could help reduce the use of lab mice in medical research. It also opens up the potential for this technique to be used in patients by quickly and confidently identifying if drugs are having a specific effect on their tumors.”
The study was published in Bioinformatics.
