Systems and methods for uncertainty aware caliper placement
Systems and methods for characterizing uncertainty on a boundary of a region of interest includes inputting, via a processor, an ultrasound image having the region of interest into a trained neural network. Systems and methods also include outputting, via the processor, from the trained neural network a first prediction of the boundary of the region of interest and a second prediction of a region segmentation of the region of interest. Systems and methods further include determining, via the processor, an uncertainty on the boundary based on mismatch between the first prediction and the second prediction.
1 . A computer-implemented method for characterizing uncertainty on a boundary of a region of interest, comprising:
inputting, via a processor, an ultrasound image having the region of interest into a trained neural network;
outputting, via the processor, from the trained neural network a first prediction of the boundary of the region of interest and a second prediction of a region segmentation of the region of interest;
determining, via the processor, an uncertainty on the boundary based on mismatch between the first prediction and the second prediction; and
automatically placing, via the processor, a pair of calipers on the region of interest in the ultrasound image in an absence of uncertainty on the boundary.
2 . The computer-implemented method of claim 1 , further comprising providing, via the processor, a user-perceptible indication of the uncertainty when mismatch is present between the first prediction and the second prediction.
3 . The computer-implemented method of claim 2 , wherein providing the user-perceptible indication of the uncertainty comprises indicating on the ultrasound image any region of uncertainty on the boundary.
4 . The computer-implemented method of claim 3 , wherein the user-perceptible indication comprises highlighting the region of uncertainty on the ultrasound image.
5 . A system for characterizing uncertainty on a boundary of a region of interest, comprising:
a memory encoding processor-executable routines; and
a processor configured to access the memory and to execute the processor-executable routines, wherein the processor-executable routines, when executed by the processor, cause the processor to:
input an ultrasound image having the region of interest into a trained neural network, wherein the ultrasound image comprises a pair of calipers marked on the region of interest;
output from the trained neural network a first prediction of the boundary of the region of interest and a second prediction of a region segmentation of the region of interest;
determine an uncertainty on the boundary based on mismatch between the first prediction and the second prediction; and
providing an alert to a user when at least a portion of the pair of calipers is placed in a region of uncertainty on the boundary.
6 . The system of claim 5 , wherein the processor-executable routines, when executed by the processor, further cause the processor to provide a user-perceptible indication of the uncertainty when mismatch is present between the first prediction and the second prediction.
7 . The system of claim 6 , wherein providing the user-perceptible indication of the uncertainty comprises indicating on the ultrasound image any region of uncertainty on the boundary.
8 . The system of claim 5 , wherein the processor-executable routines, when executed by the processor, further cause the processor to automatically place a pair of calipers on the region of interest in the ultrasound image in an absence of uncertainty on the boundary.
9 . The system of claim 5 , wherein the processor-executable routines, when executed by the processor, further cause the processor to provide a user-perceptible indication on the ultrasound image of any end of the pair of calipers placed in the region of uncertainty on the boundary.
10 . The system of claim 5 , wherein the processor-executable routines, when executed by the processor, further cause the processor to alter placement of the pairs of calipers on the region of interest in the ultrasound image when at least the portion of the pair of calipers is placed in the region of uncertainty on the boundary.
11 . A non-transitory computer-readable medium, the computer-readable medium comprising processor-executable code that when executed by a processor, causes the processor to:
input a medical image having a region of interest into a trained neural network, wherein the ultrasound image comprises a pair of calipers marked on the region of interest;
output from the trained neural network a first prediction of a boundary of the region of interest and a second prediction of a region segmentation of the region of interest;
determine an uncertainty on the boundary based on mismatch between the first prediction and the second prediction; and
alter placement of the pairs of calipers on the region of interest in the ultrasound image when at least a portion of the pair of calipers is placed in a region of uncertainty on the boundary.
12 . The non-transitory computer-readable medium of claim 11 , wherein the processor-executable code, when executed by the processor, further causes the processor to provide a user-perceptible indication of the uncertainty when mismatch is present between the first prediction and the second prediction.