Deep Learning Algorithm can detect colon cancer mutations

Deep Learning, a subset of Machine Learning, which again falls under the category of Artificial Intelligence contains an inherent potential in itself to induce recent technological advancements in the Healthcare realm.

 

Such are the ingenious and pioneering capabilities of Deep Learning that they do not just embellish the process, but also further it for the forthcoming generation to propel it towards progress. 

 

(Must Check: Emerging Technologies Revolutionizing Healthcare in 2022)

 

The Approach

 

A fresh deep learning technique developed by University of Warwick researchers can more efficiently determine the molecular pathways and development of key mutations that cause colorectal cancer than current approaches, allowing patients to benefit from therapeutic targets with faster turnaround times and lower expenses.

 

The University of Warwick's Computer Science Department has devised a machine-learning approach for predicting the genesis and mutations of colorectal cancer. 

 

(Also Check: AI in Cancer Detection and Treatment)

 

The approach eliminates the need for manual annotations on digital photos of malignant tissue slides, enabling faster and more efficient therapy.

 

As an alternative to conventional testing regimens for these processes and mutations, researchers at the Warwick University investigated how machine learning can identify three critical mutations using whole-slide imagery of Colorectal cancer slides stained with Hematoxylin and Eosin.

 

 Their findings were published in The Lancet Digital Health on 19th October.A group of researchers has developed a new method for identifying relevant sub-images or tiles from a whole-slide imagery without the need for extensive annotations by a pathologist at the cell or regional level. 

 

Basically, the new system can identify clinically significant mutations and pathways for colon cancer using raw pixel data. It works by identifying picture patches that are the best prognostic of important molecular factors in colorectal tumors using a deep convolutional neural network.

 

Researchers investigated the accuracy of iterative draw-and-rank sampling and discovered that their deep learning algorithm was far more accurate than existing published approaches for predicting the 3 primary colorectal cancer molecular pathways and major mutations.

 

(Related Reading: Other Deep Learning-based Algorithms)

 

Benefits

 

The Director of the TIA Centre at Warwick and senior author of the paper, Professor Nasir Rajpoot, stated that the work highlights how advanced algorithms can harness the potential of raw pixel data for identifying clinically relevant mutations and pathways for colon cancer. 

 

The fact that the iterative draw-and-rank sampling procedure does not require time-consuming and arduous annotations from experienced pathologists is a significant benefit.

 

As reported by NewsMedicalLifeSciences, the findings of the method suggest that repeated draw-and-rank sampling might be used to identify patients who are likely to gain from targeted treatments at lesser prices and with faster turnaround times than other special marker-based methods. 

 

To pave the path for clinical trials, the researchers will perform a substantial multi-centric validation of their method.