Computer vision is a branch of artificial intelligence that enables algorithms to extract accurate and valuable information from photos and videos. Cancer researchers have found innovative ways to use this technology in the analysis of medical images, microscopic samples, and radiological examinations, which have helped simplify complex procedures and speed up work, especially for teams with limited resources, which reflects positively on the lives of patients.
Below, we will mention the most important ways in which computer vision contributes to facilitating scientific cancer research:
1-better understanding of tumor growth factors
Usually, after checking the presence of cancer and determining its type through biopsies, pathologists resort to RNA analysis to detect genetic changes affecting tumor growth. These data are of great importance in supporting research and the development of personalized treatment plans, but traditional methods of extracting them are expensive and slow.
To address this, researchers have created an artificial intelligence tool that can analyze conventional microscopic images to predict gene activity inside cancer cells. The tool was trained on more than 7,500 samples for 16 types of cancer, as well as data for healthy cells.
The tool displays the results in the form of a visual map showing the differences between tumor areas, and has been able to determine the gene expression of more than 15 thousand genes with an accuracy of up to 80%.
2-help in choosing the most effective treatments
Cancer patients often face mental and physical suffering, especially if the initial treatments did not bring the desired results. Therefore, the ability to choose the optimal treatment early is a big difference.
Treatment plans are usually based on CT or MRI images, which provide limited data. But modern artificial intelligence technologies can analyze most of the details in the image, and even examine very fine microscopic samples, which helps in determining the appropriate treatment in the initial stages.
Using samples for bladder cancer, an artificial intelligence tool was able to identify cell populations that have a better response to immunotherapy (a type of therapy used for cancer), and also helped to assess the spread of stomach cancer with high accuracy. Scientists hope that these results will contribute to accelerating research and improving doctors ' treatment decisions.
3-accelerate the development of cancer drugs
It usually takes several years to produce a new cancer drug, and to speed up this process, a research team in London has developed an artificial intelligence tool to track the arrival of drugs to their intracellular targets.
The algorithms of the tool have analyzed more than 100 thousand three-dimensional microscopic images of skin cancer, which made it possible to observe the shape of living cells and monitor their change under the influence of treatment.
The tool has achieved more than 99% accuracy in determining the effect of drugs, with the ability to shorten preclinical development stages from three years to three months, and reduce the duration of trials by up to six years.
4-development of Comprehensive Cancer assessment models
The majority of current AI tools help in only one aspect of the work, so some research teams have developed comprehensive ChatGPT-like systems capable of performing multiple assessments of different types of cancer.
One of these systems was trained on 15 million images, 60 thousand slides of tumors of 19 different types, and data from 24 hospitals worldwide. This system analyzes digital slides and molecular files, predicts the results of treatment, and reveals the correlation of certain characteristics of tumors with increased survival rates.
5-improve the analysis of microscopic images
Traditional methods of analyzing microscopic images are time-consuming, especially if the images are large. But a new smart tool based on computer vision and machine learning algorithms can analyze samples and detect common features among cancerous tumors quickly and with high accuracy.
The instrument examines multiple areas of the tumor and treats them as an integrated unit. Other tools for analyzing microscopic images divide large tumors into small sections and treat each part as a separate sample, which makes it more difficult to study them.
In practical tests, this tool has surpassed the best current tools by nearly 4% and achieved an accuracy of up to 88%, making it applicable to any type of tumor and any imaging technique
AI-powered computer vision applications demonstrate significant potential to enhance the speed and accuracy of cancer treatment research, support doctors in diagnosis and treatment, and expedite drug development. But even with such potential, the human component remains the basis in medical decision-making, and these technologies are just supporting tools and not a substitute for doctors.