Deep Learning-Based TEM Image Analysis for Fully Automated Detection of Gold Nanoparticles Internalized Within Tumor Cell
Problem
Detecting gold nanoparticles inside tumor cells in TEM images is manual, slow, and error-prone with thresholding methods.
Why it matters
Quantifying nanoparticle uptake in tumor cells matters for nanomedicine research and cancer therapy development.
What the AI did
Deep learning models (YOLOv5 variants with transfer learning and tuned inputs) fully automated detection of gold nanoparticles from cellular TEM images.
Results
Maximum F1 score of 0.982 achieved by the best-trained models; mean average precision also reported but abstract truncated.
Impact
The automated detector achieved near-perfect F1 on validation, replacing manual analysis per the abstract.
Record
| Venue | Microscopy and Microanalysis |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Amrit Kaphle, Sandun Jayarathna, Hem Moktan, Maureen L. Aliru, Subhiksha Raghuram, Sunil Krishnan, Sang Hyun Cho |
| Citations | 16 (OpenAlex, at import time) |
| References | 62 |
| Model or method | YOLO v5 |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/micmic/ozad066 |
| Open access | green |
| Keywords | deep learning, TEM, gold nanoparticles, YOLOv5, tumor cells, automated detection |
Importance rating
real but narrower advance
Abstract states the tool will benefit R&D on GNP-based cancer therapeutics; a research-enabling imaging advance, narrower than direct clinical or industry-scale impact.