Ph.D. Student at Durr Lab, BMEResearch interests include Computational Biophotonics, and colonoscopy image analysis using deep learning.MSE Thesis at Durr Lab, BME- Using semi-supervised deep learning to improve the performance of colonoscopy image analysis. The model also gives higher performance for domain adaptation & out-of-distribution detection tasks as compared to a purely supervised model.- Dataset for 3D reconstruction of colon: Created a ground truth dataset for benchmarking depth estimation and SLAM algorithms applied to endoscopy. This involved collecting & creating silicone and tissue phantoms, recording endoscopic video sequences and developing a 2D-3D image-to-model registration algorithm.- M. Golhar, T. L. Bobrow, M. P. Khoshknab, S. Jit, S. Ngamruengphong and N. J. Durr, "Improving Colonoscopy Lesion Classification Using Semi-Supervised Deep Learning," in IEEE Access, vol. 9, pp. 631-640, 2021, doi: 10.1109/ACCESS.2020.3047544.

Initiating and inviting industry speakers and/or alumni speakers for the EDGE talk series. EDGE talk series enables JHU BME Ph.D. students to network and learn about career development opportunities from industry leaders.

Medical Imaging R&D Team - Health and Medical Equipment Division

"Blood vessel extraction in endoscopic images"- Classified endoscopic images into blood vesselcontaining and non-blood vessel containing classes using SVM. Further, performance of blood vesseldetection using Frangi Vesselness method's was improved by using custom symmetry detectionfiltering and background removal.

1) Automated detection of Choroid-Sclera Boundary in 2D-OCT Choroidal images : Developedimage segmentation algorithm based on gradient methods like edge detection andmorphological operations.2)Objective Performance evaluation of blood vessel detection algorithms for 2D choroidal OCTimages : Four similarity measures based on Dice Coefficient, Hamming distance, mismatcharea, van Dongen distance and pairs of pixels classification were proposed for comparingvarious vessel detection algorithm’s results.