a memory to store instructions;

a processor to execute the instructions stored in the memory;

wherein the system is specially configured to:

execute the instructions via the processor for receiving an input medical image of a patient, the input medical image comprising a plurality of pixels;

execute the instructions via the processor for training a generator of a Generative Adversarial Network (GAN) to detect a diseased region in the medical image of the patient via the following learning operations:

(i) performing a cross-domain translation learning operation, wherein the generator of the GAN learns to change only a portion of the plurality of pixels representing a diseased region when present in the input medical image, and not change any other pixels in the input medical image, to match a distribution of pixels in a target domain represented by un-paired training images, and

(ii) performing a same-domain translation learning operation, wherein the generator of the GAN learns to not change any pixels in the input medical image when a diseased region is absent in the input medical image so that the pixels in the input medical image match the distribution of pixels in the target domain represented by the un-paired training images;

execute the instructions via the processor for computing losses for the training of the generator of the GAN;

execute the instructions via the processor for updating a plurality of weights of the generator of the GAN; and

execute the instructions via the processor for iteratively repeating, for multiple input medical images, performing the cross-domain translation learning operation, performing the same-domain translation learning operation, computing losses for the training of the generator of the GAN, and updating the plurality of weights of the generator of the GAN until a training threshold is attained.