On December 12th, at the 2021 annual quality evaluation summary meeting of Chongqing Hematology Medical Quality Control, the "man vs. machine" battle of the DFang Cup Hematology Morphology Knowledge Competition was also held. The AI medical imaging diagnosis system, which had been studied at Xinqiao Hospital for three years, demonstrated astonishing strength and won the championship with a score of identifying 25 diseased cells in 20 seconds. Meanwhile, the experts who also achieved full marks The fastest it took was 71 seconds.

Before the competition began, the staff prepared 13 pictures of diseased cells in advance. The contestants scanned the codes on their mobile phones to enter the competition page.

Twenty seconds later, the AI medical imaging diagnosis system identified 25 diseased cells, all of which were correct, achieving a full score. The first to submit the paper was Ru Jinwei from Lechang People's Hospital, who scored a full mark in 71 seconds. Contestants from Chongqing followed closely behind. Those from Chongqing Cancer Hospital, the Second Affiliated Hospital of Chongqing Medical University, and Yubei District People's Hospital also achieved full marks one after another. In the end, a total of 32 contestants achieved full marks.

It is reported that due to issues such as the cumbersome process, strong subjectivity, and great difficulty in talent cultivation, cytological examination often requires highly experienced doctors to operate and supervise. Due to the shortage of morphological experts, many of them are mainly concentrated in tertiary hospitals, while there is a severe shortage of experts in grassroots hospitals. For some patients, the lack of diagnostic techniques may, to a certain extent, delay the best treatment opportunity.

With the advancement of smart medical technology, the AI medical imaging diagnosis system has emerged. This system specifically applies artificial intelligence technology to the diagnosis of medical images. At present, it mainly consists of two parts: the first is image recognition, which is applied in the perception stage. Its main purpose is to analyze non-structured data such as images and obtain some meaningful information. The second is deep learning, which is applied in the learning and analysis stages and is the most core link in AI applications. Through a large amount of image data and diagnostic data, deep learning training is continuously conducted on neural Service Networks to enable them to master the ability of "diagnosis".

It is equivalent to liberating doctors from repetitive work, allowing them to have more time and energy to serve patients in other aspects, and also enabling the value of doctors to be better reflected. Peng Xiangui, a senior technician from the Hematology Department of Xinqiao Hospital, said that his daily work involves examining each bone marrow smear under a microscope and issuing test reports for patients based on the different cell shapes observed with the naked eye. The AI medical imaging diagnosis system can help doctors identify each cell in the patient's bone marrow smear.

Liu Zhi, a relevant person in charge of Chongqing Defang Information Technology Co., LTD., the research and development unit of this AI medical imaging diagnosis system, introduced

The recognition speed of this AI system can reach 0.0063 seconds per cell, which can provide a solution for the shortage of morphologists in more grassroots hospitals. (Source: Chongqing Daily Client