Abhishek Ambast

Ai Engineer @AIRA Matrix

Thane, IN

Worked on multiple projects which combines medical imaging and AI, with a focus on cancer diagnostics and model optimization. Key contributions include-3D reconstruction of Prostate Cancer: Developed an algorithm to convert the 2D based RP images of prostate to 3D model which can show the approximate positioning of Lesions with grades, ISUP scores and cribriform- PD-L1 Lung Cancer Analysis: Enhanced tumor detection pipeline to improve PD-L1 TPS scoring accuracy, significantly reducing false negatives- MRI Prostate Cancer Pipeline: Tuned and automated the MRI Histopathology registration pipeline for radical prostatectomy (RP) cases, integrating heatmap visualization for improved MRI registration- MRI Lesion Detection: Developed nnUNet and SegFormer-based pipeline for MRI lesion detection. SegFormer based model focused on T2-weighted MRI, achieved higher sensitivity in lesion detection compared to multi-modal baselines- Tumor Feature Analysis: Conducted attention-based analysis of cribriform vs Tumor regions using registered T2-weighted models- Prostate Cancer Grades feature analysis: Trained a DINOv2 self-supervised vision model to extract region-level features across different prostate cancer grades, aiding in understanding of grade-specific representations- Annotation & Visualization: Supported region-level annotation workflows and 3D visualization- Monkey Kidney Biopsy GrandChallenge: Developed an inflammatory cell detection model for the Monkey Challenge, contributing to high-performing solutions and achieved rank 3 in all the leaderboards- Workflow Automation: Streamlined pipelines through automation and scripting to enhance reproducibility and efficiency.

Indian Institute of Technology, Guwahati

Bachelor of Technology - BTech

Hope Hall Foundation School

Intermediate