Effect of source spectral shape on task-based assessment of detection and resolution in optical coherence tomography.

We demonstrate the effect of the spectral shape of broadband light sources in a task-based approach for assessment of signal detection and resolution in optical coherence tomography. We define two binary tasks: The signal is either present or absent and the signal can be either resolved or not. In a transparent sample bounded by two uniform interfaces we study the minimum detectable change in the index of refraction as well as the minimum resolvable distance between the layers in correlation with the source spectral shape and power. Results show that the area under the receiver operating curve (AUC) for a signal-detection task is not affected by the shape of the spectrum but solely by its optical power, whereas spectral shaping has an effect, which we quantify, on the AUC for the resolution task. Moreover, the AUC is demonstrated in relation to the concept of system sensitivity for a signal-detection task.