January 18, 2020 - by Parul Saini, Webmedy team
Cancer is the deathly disease of all, it doesn't matter what type of malignancy it is. In 2018 alone, 9.6 million people have died worldwide because of cancer.
With the diagnosis of more than 1.7 million new cases and cancer deaths in 2019 being more than 606,000, the advancement in diagnostic methods and the process of detection of cancer is the need of the hour.
That if used properly, AI as a technology can do wonders in early detection of cancer as key to successful cancer treatment is catching it early.
All the doctors and researchers agree that if cancer is detected at an early stage, it tremendously increases the patient's chances of survival. Sadly, most cancer patients are diagnosed in the later stages of the disease. Alas, Artificial Intelligence and deep learning bring significance in solving this problem.
Artificial intelligence and deep learning have been there for long, affecting our day-to-day life by changing the way we live, learn and earn. Now doctors are showing keen interest in using AI for diagnosis, management and better therapeutic options for various diseases, notably cancer. With each passing day, the application of Artificial Intelligence is booming for the enhancement of healthcare.
Big thanks to technological advances in areas like genetics, imaging, cancer is now more likely to be caught at an earlier stage than it was decades ago.
Though, the accuracy in medical imaging diagnosis is still low, with the professionals witnessing 20-30 percent wrong negatives in chest X-rays and mammographies. AI can prevent this, and the fact that healthcare is data-rich is an added benefit. The more data visible to them, the more likely they can uncover the hidden patterns inside it that can be used to perform diagnosis.
Over time, many machine learning algorithms have been introduced, but traditional forms, like logistic regression, have demonstrated the most usefulness in clinical oncology research. 'Early detection' is one of the biggest challenges that we wish to address first when it comes to treating cancer. It is accepted that more than 80% of breast, ovarian, prostate, lung cancer are avoidable if detected early.
The neural networks are the simplest form of artificial intelligence. Machine learning is the branch of AI that is concentrated on training machines to be better at repetitive tasks. Through algorithms, that can guide systems to decide where they are right and when they are wrong, in a short period system could learn propagation rate of data. Analytical durability of the technique and usage of complex algorithms can identify patterns and behaviors. AI technology is now able to offer human-like vision and determinations needed to excel in the health sector.
Detecting cancer might be AI's big-hearted and complex challenge yet. Typical standard procedures like screening, radiological imaging can make a mistake or can return false negatives. AI will not only improve the accuracy of image detection but could also bridge the gap between cancer screening and genomics. The New York Genome Center depends on unique software for screening patients - Watson, an artificial intelligence system developed by IBM.
Sophia Genetics, in Switzerland, is using artificial intelligence to help doctors pinpoint gene mutation behind cancer. Currently, according to the company, it is being used over 80 countries by more than 970 hospitals and its costs on average $50-$200 per genetic evaluation.
Freenome for early cancer detection, another deep-tech innovation has impressed well known VCs including Google Verily. Recently, a US-based company declared strategic association with the Institut Curie to measure its AI genomics platform to predict how patients respond to immuno-oncology therapies by keeping an eye on changes in biomarkers moving in the bloodstream.
Ultrasound elastography is an almost new diagnostic technique that analyzes the stiffness of breast tissue. It accomplishes this vibrating the tissue, which generates a wave. This wave introduces exaggeration in the ultrasound scans and highlights the area of the breast where property differs. This is useful to determine if a lesion is cancerous or benign. As cancer causes different types of changes in the tissues, the existence of cancer can lead to an adjustment in its physical properties, for example, change in porosity. The role of artificial intelligence is to ascertain whether a given tissue is cancerous or not?
One of the significant capabilities of AI is to process vast and complex information in a short period. With proper teaching, AI is capable to mimic a human mind, and as it continuously advancing, AI can see more accurately, what naked human eye can't, but with even greater efficiency.
Typically, using such an approach is less invasive and traumatic for patients and can also speed up the diagnosis process, making sure treatment starts sooner.
There is nothing new to say that industry has a long way to go and AI needs to beat its trust issues and to prove itself to patients and doctors. In my opinion, the in-built lack of trust is the biggest obstacle in full adoption. AI alone should not be expected as a solution for diagnosis and treating a patient in a fully unguided environment. Fairly, it's a smart solution and helpful assistance to doctors, who have a holistic view, of the patient's situation and medical history.
It's a good example of how technology and humans can work in-sync, not against one another. AI as what is capable of doing: that is going to help in fields where human skills remain limited.
What I think about the future and I wish cancer diagnosis will look like, I'd like to watch AI set up to accelerate diagnosis and methods, even recovery. But generally, I'd like to watch technology getting more easily available and used to boost survival rates, improve treatments and minimize side effects.