Chinese scientists have developed a learning artificial intelligence (AI)
A system that can accurately diagnose and identify cancer prostate samples like any pathologist.
This provides the possibility to simplify and eliminate changes in the course of cancer diagnosis.
It also helps to overcome the lack of trained pathologists in the local area.
In the long run, it can lead to automation or part-
Automatic diagnosis of prostate cancer.
Prostate cancer is the most common cancer in men, about 1.
1 million cases are diagnosed each year.
A biopsy sample is usually required to confirm the diagnosis, which is then examined by the pathologist.
Now, the artificial intelligence learning system presented at the conference of the European Urology Association in Copenhagen shows similar accuracy to that of human pathologists.
In addition, the software can accurately classify the malignant degree of cancer, thus eliminating the variability that may penetrate into human diagnosis.
"This will not replace human pathologists.
We also need an experienced pathologist to be responsible for the final diagnosis.
It will help pathologists make better and faster diagnosis and eliminate the time of dayto-
Guo Hongqian of Nanjing University, China, said: "Changes in daytime judgment may penetrate into human assessment . ".
The researchers collected 918 samples of the whole prostate pathological slice from 283 patients, and analyzed them through the analysis system. the software gradually learned and improved the diagnosis level.
These pathological images were divided into 40,000 smaller samples;
30,000 of these samples were used for "training" software and the remaining 10,000 for testing accuracy.
The results showed that 99 cases were diagnosed accurately.
38 of the cases (
Using human pathologists as the "Golden Standard ")
As accurate as a human pathologist.
"The system is programmed to learn and gradually improve the way it explains the samples.
Our results show that the level of diagnosis reported by AI is comparable to that reported by pathologists, "Guo said.
"In addition, it can accurately classify the malignant degree of prostate cancer," Guo added . ".