Diagnostic model optimization for variations in mechanical components of imaging devices
Introduced here are diagnostic platforms able to optimize computer-aided diagnostic (CADx) models by simulating the optical performance of an imaging device based on its mechanical components. For example, a diagnostic platform may acquire a source image associated with a confirmed diagnosis of a medical condition, simulate optical performance based on design data corresponding to a virtual prototype of the imaging device, generate a training image by altering the source image based on the optical performance, apply a diagnostic model to the training image, and then determine whether the performance of the diagnostic model meets a specified performance threshold. If the diagnostic model fails to meet the specified performance threshold, the diagnostic platform can automatically optimize the diagnostic model for the imaging device by altering its underlying algorithm(s).
1. An electronic device comprising:
a memory that includes instructions for validating a diagnostic model developed to facilitate computer-aided diagnosis of a medical condition,
wherein the instructions, when executed by a processor, cause the processor to:
acquire a first set of images generated by a first imaging device;
identify a second imaging device for which the diagnostic model is to be optimized;
acquire a second set of images generated by the second imaging device;
apply a translation algorithm to discover a translation scheme that specifies how to translate a first image in the first set of images into a second image in the second set of images,
wherein the first and second images are associated with a same subject; and
optimize the diagnostic model for the second imaging device based on the translation scheme.
2. The electronic device of claim 1 , wherein the instructions further cause the processor to:
save the optimized diagnostic model in the memory; and
associate the optimized diagnostic model with the second imaging device in a database record.
3. The electronic device of claim 1 , wherein the translation scheme specifies what operations, if any, are necessary to make the first and second images appear substantially identical.
4. The electronic device of claim 1 , wherein the second imaging device is a proposed imaging device embodied as a virtual prototype.
5. The electronic device of claim 4 , wherein the instructions further cause the processor to:
access design data related to the virtual prototype of the second imaging device;
predict optical performance of the second imaging device by simulating the transmission of light through the second imaging device during an imaging operation using the design data; and
generate the second set of images by modifying each image in the first set of images based on the predicted optical performance.
6. The electronic device of claim 2 , wherein said associating causes the optimized diagnostic model to be programmatically linked to the second imaging device, such that the optimized diagnostic model is applied to images generated by the second imaging device in the future.
7. The electronic device of claim 5 , wherein the instructions further cause the processor to:
parse the design data to identify a set of hardware components through which the light will extend during the imaging operation.
8. The electronic device of claim 7 , wherein said predicting comprises:
transmitting the design data related to the virtual prototype of the second imaging device to a computer program that is designed to simulate the transmission of light through the set of hardware components, and
receiving, from the computer program, output indicative of the predicted optimal performance.
9. The electronic device of claim 1 , wherein the translation scheme involves a change in blurriness, contrast, white balance, sharing, scattering, distortion, color aberration, sharpening, compression, or any combination thereof.
10. The electronic device of claim 1 , wherein the diagnostic model is a classification model that is designed to employ one or more statistical techniques to pixel data corresponding to a given image provided as input, so as to produce output that is representative of a proposed diagnosis for the medical condition.