PARIS, April 6 (Benin News) –
Using a fully automated deep learning model based on artificial intelligence (AI), researchers were able to identify early signs of type 2 diabetes in abdominal CT scans, according to a new study published in the journal “Radiology”.
Due to the slow onset of symptoms, it is important to diagnose type 2 diabetes early. Some cases of pre-diabetes can last up to eight years and early diagnosis will allow patients to make lifestyle changes to alter the progression of the disease.
Abdominal CT could be a promising tool for the diagnosis of type 2 diabetes. CT is already widely used in clinical practice and can provide a significant amount of information about the pancreas.
Previous studies have shown that diabetic patients tend to accumulate more visceral fat and fat in the pancreas than non-diabetic patients. However, little work has been done to study the liver, muscles and blood vessels surrounding the pancreas, says study co-author Ronald M. Summers, principal investigator and radiologist at the National Institutes of Health Clinical Center in Bethesda. , in the USA.
“The analysis of both pancreatic and extrapancreatic characteristics is a novel approach that, to our knowledge, has not been demonstrated in previous work,” adds the student and first author.
Manual analysis of low-dose non-contrast pancreatic CT images by a trained radiologist or specialist is a time-consuming and difficult process. To address these clinical challenges, automated pancreatic image analysis needs to be improved, the authors note.
For this retrospective study, Dr. Summers and colleagues, working closely with co-author Perry J. Pickhardt, professor of radiology at the University of Wisconsin School of Medicine and Public Health, used a dataset on patients who had undergone routine screening for colorectal cancer by computed tomography at the hospital and clinics of the University of Wisconsin.
Of the 8,992 patients who were screened between 2004 and 2016, 572 were diagnosed with type 2 diabetes and 1,880 with dysglycemia, a term for blood sugar levels that are too low or too high. There was no overlap between the diagnosis of diabetes and that of dysglycemia.
To build the deep learning model, the researchers used a total of 471 images obtained from various datasets, including the Medical Data Decathlon, the Cancer Image Archive, and the Beyond Cranial Vault challenge. The 471 images were divided into three subsets: 424 for training, eight for validation, and 39 for testing. The researchers also included data from four sets of active learning.
The deep learning model showed excellent results, demonstrating almost no difference compared to manual analysis. Besides the different characteristics of the pancreas, the model also analyzed visceral fat, density and volumes of surrounding abdominal muscles and organs.
The results showed that diabetic patients had lower pancreatic density and a higher amount of visceral fat than non-diabetic patients.
“We found that diabetes was associated with the amount of fat inside the patients’ pancreas and abdomen,” Dr. Summers points out. The more fat there was in these two places, the more patients were likely to suffer from diabetes for a longer period of time.
The best predictors of type 2 diabetes in the final model included intrapancreatic fat percentage, pancreatic fractal dimension, plaque severity between the level of L1-L4 vertebrae, average liver CT attenuation, and BMI. The deep learning model used these predictors to accurately discern patients with and without diabetes.
“This study is a step towards wider use of automated methods to address clinical challenges,” the authors state. “It can also serve as a basis for future work. It can also serve as a basis for future work to determine the reason for pancreatic changes in diabetic patients.
We would like to give thanks to the writer of this article for this outstanding material
Artificial intelligence can improve the diagnosis of diabetes – Benin Actu
Find here our social media profiles as well as other pages related to it.https://www.ai-magazine.com/related-pages/