European Radiology: Lung-RADS classification application of computer-aided diagnosis in CT screening of lung cancer

At this stage , low-dose CT for lung cancer screening has been more and more clinically accepted and widely used in clinical practice
.

In order to standardize the classification and description of pulmonary nodules , the Lung Imaging Reporting and Data System (Lung-RADS) was introduced clinically to assist radiologists in the interpretation of pulmonary nodules .

At this stage , lung cancer screening with low -dose CT has been increasingly accepted in clinical practice and widely used in clinical practice .
In order to standardize the classification and description of pulmonary nodules , the Lung Imaging Reporting and Data System (Lung-RADS) was introduced clinically to assist radiologists in the interpretation of pulmonary nodules .
The Lung-RADS classification is primarily determined by nodule size, nodule type, and growth, and each assessed aspect produces inter-reader variability .
A onlyprevious moderate (mean Cohen's kappa, 0.
51) .

To reduce inter-reader variability and discriminative differences , Lung- RADS classification of lung nodules can be performed using an automated segmentation tool .
Recent advances in deep learning applications in medical imaging go beyond simple nodule detection to enable automatic segmentation, classification, and measurement of nodules, and assessment of their malignancy risk , a system known as computer-aided diagnosis (CAD) .
diagnosis

Recently, a study published in the journal European Radiology evaluated the impact of CAD on inter-reader consistency in pulmonary RADS classification, providing a guarantee for the automatic detection and risk assessment of pulmonary nodules, and improving the consistency of image interpretation.

sex provides a reference .
 

Recently, a study published in the journal European Radiology evaluated the impact of CAD on inter-reader consistency in pulmonary RADS classification, providing a guarantee for the automatic detection and risk assessment of pulmonary nodules, and improving the consistency of image interpretation.
sex provides a reference .
 

200 baseline CT scans covering all lung Lung-RADS categories were randomly selected from the National Lung Cancer Screening Trial

Five readers reported 139-151 negative screening results without CAD and 126-142 results with CAD

 

A, b Axial and coronal CT images show a fibrous nodule of approximately 18.

8 mm in the left upper lobe .
In the first evaluation , only reader 4 selected this lesion as a risk-dominant nodule and classified it as a pulmonary RADS category 4B .
Readers 1 and 2 selected other small nodules as risk-dominant nodules (c, d), and readers 3 and 5 did not select any nodules .
Since CAD selected the same lesions as Reader 4, Readers 1 and 5 followed the CAD results .

FIG.
A, b-axis position and coronal bit CT image shows the left upper lobe about 18.
8 mm fiber nodules .
In the first evaluation , only reader 4 selected this lesion as a risk-dominant nodule and classified it as a pulmonary RADS category 4B .
Readers 1 and 2 selected other small nodules as risk-dominant nodules (c, d), and readers 3 and 5 did not select any nodules .
Since CAD selected the same lesions as Reader 4, Readers 1 and 5 followed the CAD results .
However, Reader 2 selected the same nodules as in the 1st assessment , even with the markers for CAD , while Reader 3 still did not select any nodules as risk-dominant nodules .
Based on these results, we speculate that some readers do not consider this fibrous nodule to constitute a "nodule," although one reader considered this lesion to be a nodule in the absence of CAD findings .

Taken together, the application of CAD demonstrated improved inter-reader agreement in the Lung-RADS category , while showing a reduction in measurement variability in lung cancer -positive cases , and could further improve clinical recognition of pulmonary nodules .
Standardization of diagnosis and treatment .

Original source :

Sohee Park , Hyunho Park , Sang Min Lee ,et al .

Application of computer-aided diagnosis for Lung-RADS categorization in CT screening for lung cancer: effect on inter-reader agreement

Sohee Park Hyunho Park Sang Min Lee ,et al Application of computer-aided diagnosis for Lung-RADS categorization in CT screening for lung cancer: effect on inter-reader agreement

.

DOI : 10.
1007/s00330-021-08202-3

10.
1007/s00330-021-08202-3 10.
1007/s00330-021-08202-3 Leave a message here