Applications: IVF

This work presents a label-free, high-resolution optical imaging platform that combines quantitative phase imaging (QPI) with advanced machine learning to assess the quality of human sperm cells under various stress conditions relevant to assisted reproductive technologies (ART). Conventional bright-field microscopy, while widely used for sperm selection during procedures like ICSI (Intracytoplasmic Sperm Injection), lacks the sensitivity to detect subcellular structural changes that impact fertility. To address this limitation, a partially spatially coherent digital holographic microscopy (PSC-DHM) system was developed to enable highly sensitive QPI of live sperm cells. This imaging capability enables the extraction of fine morphological features and dynamic cellular responses without labeling or staining. Sperm cells subjected to stress conditions, such as cryopreservation, oxidative stress via hydrogen peroxide, and ethanol exposure, were imaged and analyzed to investigate changes in morphology and biophysical parameters.

In a large-scale study involving over 10,000 sperm cells, deep neural networks (DNNs) and support vector machines (SVMs) were trained on the phase data to classify healthy and stressed cells. The machine learning models demonstrated high accuracy, sensitivity, and specificity, successfully identifying stress-induced changes that correlate with reduced fertility potential, such as altered cell volume, decreased phase amplitude, and morphological deformities.

References:

1. Dubey, Vishesh, Daria Popova, Azeem Ahmad, et al. "Partially spatially coherent digital holographic microscopy and machine learning for quantitative analysis of human spermatozoa under oxidative stress condition." Scientific reports 9, no. 1 (2019): 3564. Link 

2. Butola, Ankit, Daria Popova, Dilip K. Prasad, Azeem Ahmad, et al. "High spatially sensitive quantitative phase imaging assisted with deep neural network for classification of human spermatozoa under stressed condition." Scientific reports 10, no. 1 (2020): 13118. Link