Lung RSA

Analysis of existing archives of lung CT images collected during the COVID-19 pandemic for malignant neoplasms (MN) and their characteristics, LUNG-RADS recommendations

The algorithm processes CT lung images, highlighting potentially dangerous areas and drawing the specialist's attention to nodular masses. If the AI "sees" a neoplasm 4 mm in size, it highlights all nodules on the image regardless of their size. The doctor then analyzes the images and makes a final decision.

How it works

How it works

1. The algorithm works with existing archives and retrospectively analyzes DICOM studies without contrast.

2. The images are processed by artificial intelligence, after processing the doctor receives images with the identification of potentially dangerous areas and text description of the study.

3. AI-confirmed cases are automatically sent to the MDDC Reference Center or to the regional oncological dispensary for verification by specialist physicians.

Benefits

Mass screening of the population through the use of off-the-shelf CT scans

Increasing early detection of cancer

Saving the time and efforts of radiologists

Increasing the effectiveness of treatment through early detection

Reducing the burden on the patient using heavy diagnostic equipment

Increased diagnostic speed and accuracy

Integration

Karachay-Cherkessia Oncologic Dispensary named after S.P. Butov. S.P. Butov

Karachay-Cherkessia Republican Clinical Hospital

Source data

During the pandemic in the region the "CT Lung" service was integrated to detect COVID-19 on the images.

Solution

The CT images accumulated during the pandemic were further analyzed by artificial intelligence for cancer detection.
The retrospective analysis allowed mass screening of patients without additional CT scans.
The pilot project "Retrospective analysis of pulmonary CT scans" was conducted from 01.09.2021-18.09.2021 in the Karachay-Cherkess Republic.

Results

~ 1 500

depersonalized images were analyzed for nodular neoplasms

12

patients in whom AI detected nodular neoplasms on archived images collected for COVID-19 detection

8

cases with detected neoplasms with confirmed signs of cancer

Nizhny Novgorod region
Analysis of all CT scans of the region,

to detect early signs of cancer.

Source data

The Medical Digital Diagnostic Center (MDDC) platform is directly connected to the Medical Information System (MIS) in Nizhny Novgorod Oblast. 
The central archive of medical images receives lung CT scans taken in all medical institutions of the region. The data are automatically transferred to the MDDC platform and analyzed by the AI.

Solution

Regular online analysis of chest CT scans using AI. Every week the algorithm processes about 300 studies. 
After verification by MDDC doctors, a report with detected cases of nodular neoplasms in the lungs is sent to the Nizhny Novgorod Oncological Dispensary.

Results

Nizhny Novgorod region                          
Period from April to July 2022    
MO Nizhny Novgorod Oncological Dispensary

3155

images were analyzed for the presence of nodular neoplasms

231

cases of possible neoplasms were identified by AI and sent images for additional verification to MDDC doctors

125

Lung CT scans were sent to Nizhny Novgorod Oncologic Dispensary physicians for decision-making on further routing and further examination of patients

Feedback

“The use of artificial intelligence technologies is one of the main trends in modern medicine. For example, during the COVID-19 pandemic in many regions of Russia, artificial intelligence was connected to revise CT studies of patients. However, no less important is the use of high technology in oncology. Algorithms reduce the time of medical data processing, freeing doctors from routine work. This is especially important for oncologists, because sometimes lives can be saved by timely diagnosis”

Zuber Makhov, M.D., is a senior freelance oncologist of Karachay-Cherkessia, and head doctor at the S.P. Butov Karachay-Cherkessia Oncology Center