Algorithms for mapping kidney tissue oxygenation during normothermic machine perfusion using hyperspectral imaging

Area of Science:

• Transplantation Medicine
• Biomedical Engineering
• Medical Imaging

Background:

• Organ donor shortages necessitate improved preservation and utilization of transplantable organs.
• Normothermic machine perfusion (NMP) offers a promising alternative to cold storage, enhancing organ viability and allowing physiological assessment.
• Accurate methods for evaluating organ parameters during NMP are crucial but underdeveloped.

Purpose of the Study:

• To identify a more accurate algorithm for calculating tissue oxygen saturation (StO2) in kidneys during NMP compared to previous discrete wavelength (DW) methods.
• To evaluate full spectral algorithms, specifically area under the curve and partial least square regression (PLSR), for StO2 assessment.

Main Methods:

• Hyperspectral imaging (HSI) was employed to measure StO2 in kidneys undergoing NMP.
• Three algorithms were tested: a discrete wavelength (DW) algorithm, an area under the curve algorithm, and a partial least square regression (PLSR) algorithm.
• Calibration data was acquired to train and evaluate each algorithm within the 590-800 nm wavelength range.

Main Results:

• The 590-800 nm wavelength range was identified as optimal for analyzing kidney StO2 during NMP.
• The partial least square regression (PLSR) algorithm demonstrated superior performance in accurately assessing tissue oxygen status.
• PLSR showed promising results for real-time oxygen status evaluation in perfusion experiments.

Conclusions:

• The PLSR algorithm, utilizing hyperspectral imaging, offers a more accurate method for determining kidney tissue oxygen saturation during NMP.
• This advancement supports improved organ assessment and preservation strategies in transplantation medicine.
• Further development of spectral analysis methods can enhance the utility of NMP in evaluating organ viability.