NUST Researcher Develops DxVision to Generate Virtual H&E-Stained Images, Potentially Reducing Staining Turnaround Time by 40%

Pakistan is making strides in the application of artificial intelligence to healthcare, with Dr Muhammad Usman Akram of the National University of Sciences and Technology’s (NUST) College of Electrical and Mechanical Engineering (CoEME) developing an AI-based digital staining system designed to help pathologists examine tissue samples more efficiently.

The system, called DxVision, converts images of unstained tissue samples into virtually stained images resembling those produced through conventional haematoxylin and eosin (H&E) staining. This technique could help streamline an important step in pathological examination by enabling specialists to visualise tissue structures without relying exclusively on the conventional staining process.

H&E staining is widely used in pathology to highlight tissue architecture and cellular features that help pathologists assess samples and identify abnormalities, including those associated with cancer. Conventional staining requires laboratory processing, which adds time to the diagnostic workflow.

DxVision uses artificial intelligence to generate virtual H&E-stained images from unstained tissue images. According to the information provided about the system, the technology has the potential to reduce staining turnaround time by up to 40%, potentially helping pathology laboratories process samples more efficiently.

AI and the Future of Pathology

The development highlights the growing potential of AI-assisted technologies to support medical professionals by improving laboratory workflows and reducing delays in diagnostic procedures.

If validated through further testing, digital staining technology could offer a useful addition to pathology services, particularly in settings where diagnostic capacity, laboratory resources and turnaround times are significant concerns. Faster processing could help specialists examine samples sooner and support more efficient clinical decision-making.

However, virtual staining is intended to assist pathological examination, not independently establish a cancer diagnosis. The diagnostic significance of any tissue sample must be assessed by qualified medical professionals, and the accuracy and clinical reliability of AI-generated images require rigorous evaluation.

Pre-Clinical Trials and Potential National Application

DxVision is currently in the pre-clinical trial stage, according to the information shared about the project. Its potential for wider implementation will depend on the results of validation studies, clinical performance assessments and the ability to integrate the technology into existing laboratory workflows.

If its effectiveness and safety are established, the system could have potential for adoption at a broader, potentially national, level. Further evaluation would be needed to determine how well it performs across different tissue types, laboratory conditions and diagnostic settings.

The project brings attention to Pakistan’s emerging role in applying artificial intelligence and engineering research to healthcare challenges. It also illustrates how locally developed technologies could contribute to more efficient diagnostic services, provided they meet the necessary standards for clinical accuracy and reliability.

As AI continues to advance in medical imaging and diagnostics, systems such as DxVision may help reshape parts of the pathology workflow. For now, its progress through pre-clinical evaluation will be an important step in determining whether its promising time-saving potential can translate into reliable clinical use.