Viable and necrotic tumor assessment from whole slide images of osteosarcoma using machine-learning and deep-learning models
Problem
Pathological estimation of viable versus necrotic tumor after chemotherapy in osteosarcoma requires laborious analysis of whole slide images.
Why it matters
Tumor necrosis assessment is essential for osteosarcoma patients receiving chemotherapy.
What the AI did
Thirteen machine-learning models were trained on 40 digitized slides to label viable tumor, necrotic tumor, and non-tumor; a deep-learning architecture was also trained on the same data.
Results
Models reported to perform 'exceptionally well'; numeric metrics not stated in source
Impact
Reported as first fully automated osteosarcoma necrosis assessment tool; clinical adoption not stated.
Record
| Venue | PLoS ONE |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | Harish Babu Arunachalam, Rashika Mishra, Ovidiu Daescu, Kevin Cederberg, Dinesh Rakheja, Anita L. Sengupta, David S. Leonard, Rami R. Hallac et al. |
| Citations | 146 (OpenAlex, at import time) |
| References | 46 |
| Model or method | Support Vector Machine (top performer) plus a deep-learning architecture |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1371/journal.pone.0210706 |
| Open access | gold |
| Keywords | osteosarcoma, whole slide images, tumor necrosis, machine learning, deep learning, pathology |
Importance rating
real but narrower advance
Automated necrosis assessment is essential for osteosarcoma patients per abstract, but no scale given; a real but narrow pathology advance.