Research Projects

Cervical cancer mortality remains high, especially in low-resource settings, due to limitations in current screening methods. The number of new cases of cervical cancer in India is expected to increase 1.5-fold between 2020 (1,35,929) and 2040 (2,09,169) as per the IARC GLOBOCAN Data Forecast. An early detection of cervical cancer and suitable treatment can improve the survival rate of up to 90%. This project addresses this challenge by developing a handheld polarization imaging probe using Mueller Matrix Technique for early detection of precancerous cervical lesions. Objectives: In this project the IMX250MYR CMOS imaging sensor will be used to perform a compact, cost-effective, and real-time Mueller Matrix Polarimetry measurements such as retardance, birefringence and azimuthal. A pilot clinical study span across two phase (Phase I: N = 20 and Phase 2: N = 30) will assess the probe's ability to differentiate between healthy and precancerous cervical tissue. Development of a Artificial Intelligence (AI) Assisted Graphical User Interface (GUI) to the analyse polarimetric data to enhance diagnostic accuracy and identify optimal diagnostic markers. Rationale: It has been observed in recent studies that precancerous cervical tissue exhibits distinct polarization signatures compared to healthy tissue due to the tumor-induced blood vessels possessing irregular morphology, impacting light scattering properties detectable by Mueller Matrix Polarimetry (MMP). The cellular density and nuclear size variations in cancerous tissue influence scattering, indirectly measurable with MMP. The precancerous changes disrupts the orientation and density of the collagen fiber organization, affecting tissue birefringence, a key MMP parameter. Experiments: The main experiments involve the design and development of handheld probe integrating the IMX250MYR sensor based camera and optimizing the probe optics including the lens for the cervical imaging. Conducting a study on a diverse patient population to acquire polarization images of healthy and precancerous cervical tissue. Train and validate machine learning algorithms using the acquired data to identify precancerous lesions. The aim is to provide a real-time assistance to the clinician for additional guidance for performing the excisional biopsy for the confirmation of cervical cancer. Significance: The successful completion of this project will enable earlier detection of cervical cancer, leading to timely treatment and reduced mortality. It will offer a quantitative alternative in addition to the current subjective methods such as colposcopy, further improving diagnostic accuracy and consistency. The work will also create knowledge as to how the quantifiable Mueller Matrix polarimetric properties such as retardance, birefringence and azimuthal can be correlated with the tumour biomarkers such as collagen density and orientation, scattering due to disorder growth of the cancerous cell, and angiogenesis.