Artificial intelligence has been used to brew better beer, write music,
and read your imagination, but now it is being used by Japanese
scientists to detect bowel cancer.
It is already improving the accuracy of breast cancer diagnosis in America, but the new study has shown how machine learning can red flag a suspicious-looking colon polyp, and get it right 86% of the time.
Not only that, it only takes one second to deliver the results.
There are approximately 41,000 new cases of bowel cancer diagnosed every year in the UK, according to Cancer Research UK, but that number only makes up 11% of the overall number of patients living with the disease.
Often detected late, at which point it can be fatal, and with the prognosis for 43% of patients not surviving ten years after their diagnosis, it is essential to develop more efficient methods of diagnosis.
The new endoscopic system, presented at the United European Gastroenterology (UEG) conference in Barcelona, has already been tested in a clinical setting during the routine colonoscopy of 250 men and women.
It works by using a highly-detailed magnified image- 500 times the naked eye - of a colorectal polyp (a small clump of cells that forms on the lining of the colon), some of which can develop into colon cancer.
The system then compares over 300 features of that image against a catalogue of over 30,000 endocytoscopic images allowing it to predict the lesion pathology in less than one second.
Dr Yuichi Mori from Showa University in Yokohama, said: “The most remarkable breakthrough with this system is that artificial intelligence enables real-time optical biopsy of colorectal polyps during colonoscopy, regardless of the endoscopists’ skill.”
Overall, 306 polyps were assessed real-time by using the AI-assisted system, providing a sensitivity of 94%, specificity of 79%, and accuracy of 86%.
Not only is it accurate, but it also prevents the unnecessary removal of non-cancerous polyps, according to Dr Mori.
The NHS says bowel cancer is a general term for cancer that begins in the large bowel. Depending on where the cancer starts, bowel cancer is sometimes called colon or rectal cancer.
The three main symptoms are persistent blood in the stools, change in your bowel habit and a persistent lower abdominal (stomach) pain, bloating or discomfort.
Bowel cancer is the third most common cancer in females in the UK, with around 18,400 cases diagnosed in 2014.
AI Can Detect Bowel Cancer In Less Than A Second
Incidence rates for bowel cancer in the UK are highest in people aged 85-89 and more than 44% bowel cancer cases in the UK each year are diagnosed in people aged 75 and over.
It is already improving the accuracy of breast cancer diagnosis in America, but the new study has shown how machine learning can red flag a suspicious-looking colon polyp, and get it right 86% of the time.
Not only that, it only takes one second to deliver the results.
There are approximately 41,000 new cases of bowel cancer diagnosed every year in the UK, according to Cancer Research UK, but that number only makes up 11% of the overall number of patients living with the disease.
Often detected late, at which point it can be fatal, and with the prognosis for 43% of patients not surviving ten years after their diagnosis, it is essential to develop more efficient methods of diagnosis.
The new endoscopic system, presented at the United European Gastroenterology (UEG) conference in Barcelona, has already been tested in a clinical setting during the routine colonoscopy of 250 men and women.
#UEGWeek press release: Artificial intelligence: is this the future of early colorectal cancer detection? https://t.co/tB6jnUjIzx pic.twitter.com/xc09ybFNOt
— my_UEG (@my_ueg) October 30, 2017
It works by using a highly-detailed magnified image- 500 times the naked eye - of a colorectal polyp (a small clump of cells that forms on the lining of the colon), some of which can develop into colon cancer.
The system then compares over 300 features of that image against a catalogue of over 30,000 endocytoscopic images allowing it to predict the lesion pathology in less than one second.
Dr Yuichi Mori from Showa University in Yokohama, said: “The most remarkable breakthrough with this system is that artificial intelligence enables real-time optical biopsy of colorectal polyps during colonoscopy, regardless of the endoscopists’ skill.”
Overall, 306 polyps were assessed real-time by using the AI-assisted system, providing a sensitivity of 94%, specificity of 79%, and accuracy of 86%.
Not only is it accurate, but it also prevents the unnecessary removal of non-cancerous polyps, according to Dr Mori.
The NHS says bowel cancer is a general term for cancer that begins in the large bowel. Depending on where the cancer starts, bowel cancer is sometimes called colon or rectal cancer.
The three main symptoms are persistent blood in the stools, change in your bowel habit and a persistent lower abdominal (stomach) pain, bloating or discomfort.
Bowel cancer is the third most common cancer in females in the UK, with around 18,400 cases diagnosed in 2014.
AI Can Detect Bowel Cancer In Less Than A Second
Incidence rates for bowel cancer in the UK are highest in people aged 85-89 and more than 44% bowel cancer cases in the UK each year are diagnosed in people aged 75 and over.
Comments
Post a Comment