Visualizing Confidence value with evident improvements

We explained in our last article how Artificial Intelligence is doing wonders in the healthcare industry by making diagnosis and treatments easier. We worked on a project in the line of Lung Cancer diagnosis and treatment.


The output of the algorithm is the slice position for L3 and its associated confidence value for the position. Though both images detected L3, the adjusted image has a higher confidence value in the location.

The predicted L3 slice location was the same in both instances. The max confidence for the minimum HU value (minv) of 100 was 0.27, and the max confidence for minv=200 was 0.33.



During the testing of the code, the team observed changes in the output when pre-processing parameters were changed. The results must be verified by a Subject Matter Expert ( SME), but the above example demonstrates a change in detection from “review” to “yes” based on adjusting the minimum HU parameter.

We will be summarizing the whole project, along with the lessons learned and conclusive comments from our experts, next week. Please follow us on LinkedIn to stay tuned.