Keywords:
Artificial Intelligence, Lung, Anatomy, Digital radiography, Diagnostic procedure, Infection
Authors:
L. Rusko1, A. Radics2, V. Venugopal3, K. Nye4, G. Avinash3, V. Mahajan3; 1Szeged/HU, 2Budapest/HU, 3New Delhi/IN, 4Milwaukee, WI/US
DOI:
10.26044/ecr2019/C-2327
Results
Sample Outputs:
The chest cavity segmentation closely follows the left and right lung segmentation demonstrating repeatability in both the cases (Fig.
1 and Fig.
2). With the advancements in deep learning,
pneumothorax detection and segmentation algorithms are emerging within the market. The following experiment simulates what could be possible in the future with our lung segmentation algorithms paired with pneumothorax segmentation,
by using manually segmented PTX regions as a substitute for the demonstration. The intersection of the ground truth PTX masks and the right/left lung segmentation was used to determine the outer boundary of the PTX region,
as to represent PTX regions within the automatic lung segmentation area.
Both Figure 1 and 2,
demonstrate how this method could be used to identify the appearance of a new PTX (Figure 2a & 2b),
the worsening of the PTX (Figure 2b and 2c),
and the improvement of a PTX (Figure 1a and 1b).
The method of using a lucent lung area is demonstrated as relevant in Figure 1a and 1b,
as the presence of a pleural effusion will decrease the lucent lung area or aerated lung region,
and hence the calculated ratio reflects this impact on the patient. The method of using a ratio to remove magnification errors from patient positioning or source to image distance (SID) is shown to be beneficial in Figure 2b and 2c.
Results
The network achieved an average Dice Score Co-efficient of 95.6%,
95.3%,
and 97.1% when comparing the predicted and ground-truth masks for left lung cavity,
right lung cavity and Chest Cavity respectively.
|
DICE score |
Precision |
Re-call |
Chest Cavity |
97.1% + 2.0% |
97.5% + 3.1% |
96.8% + 2.8% |
Right Lung |
95.3% + 4.5% |
95.5% + 6.4% |
95.4% + 4.0% |
Left Lung |
95.6% + 4.0% |
95.6% + 5.2% |
95.8% + 4.4% |